2022-02-09 07:41:04 -05:00
|
|
|
package hugolib
|
|
|
|
|
|
|
|
import (
|
|
|
|
"bytes"
|
2022-04-16 04:40:21 -04:00
|
|
|
"encoding/base64"
|
2023-01-04 12:24:36 -05:00
|
|
|
"errors"
|
2022-02-09 07:41:04 -05:00
|
|
|
"fmt"
|
|
|
|
"io"
|
all: Rework page store, add a dynacache, improve partial rebuilds, and some general spring cleaning
There are some breaking changes in this commit, see #11455.
Closes #11455
Closes #11549
This fixes a set of bugs (see issue list) and it is also paying some technical debt accumulated over the years. We now build with Staticcheck enabled in the CI build.
The performance should be about the same as before for regular sized Hugo sites, but it should perform and scale much better to larger data sets, as objects that uses lots of memory (e.g. rendered Markdown, big JSON files read into maps with transform.Unmarshal etc.) will now get automatically garbage collected if needed. Performance on partial rebuilds when running the server in fast render mode should be the same, but the change detection should be much more accurate.
A list of the notable new features:
* A new dependency tracker that covers (almost) all of Hugo's API and is used to do fine grained partial rebuilds when running the server.
* A new and simpler tree document store which allows fast lookups and prefix-walking in all dimensions (e.g. language) concurrently.
* You can now configure an upper memory limit allowing for much larger data sets and/or running on lower specced PCs.
We have lifted the "no resources in sub folders" restriction for branch bundles (e.g. sections).
Memory Limit
* Hugos will, by default, set aside a quarter of the total system memory, but you can set this via the OS environment variable HUGO_MEMORYLIMIT (in gigabytes). This is backed by a partitioned LRU cache used throughout Hugo. A cache that gets dynamically resized in low memory situations, allowing Go's Garbage Collector to free the memory.
New Dependency Tracker: Hugo has had a rule based coarse grained approach to server rebuilds that has worked mostly pretty well, but there have been some surprises (e.g. stale content). This is now revamped with a new dependency tracker that can quickly calculate the delta given a changed resource (e.g. a content file, template, JS file etc.). This handles transitive relations, e.g. $page -> js.Build -> JS import, or $page1.Content -> render hook -> site.GetPage -> $page2.Title, or $page1.Content -> shortcode -> partial -> site.RegularPages -> $page2.Content -> shortcode ..., and should also handle changes to aggregated values (e.g. site.Lastmod) effectively.
This covers all of Hugo's API with 2 known exceptions (a list that may not be fully exhaustive):
Changes to files loaded with template func os.ReadFile may not be handled correctly. We recommend loading resources with resources.Get
Changes to Hugo objects (e.g. Page) passed in the template context to lang.Translate may not be detected correctly. We recommend having simple i18n templates without too much data context passed in other than simple types such as strings and numbers.
Note that the cachebuster configuration (when A changes then rebuild B) works well with the above, but we recommend that you revise that configuration, as it in most situations should not be needed. One example where it is still needed is with TailwindCSS and using changes to hugo_stats.json to trigger new CSS rebuilds.
Document Store: Previously, a little simplified, we split the document store (where we store pages and resources) in a tree per language. This worked pretty well, but the structure made some operations harder than they needed to be. We have now restructured it into one Radix tree for all languages. Internally the language is considered to be a dimension of that tree, and the tree can be viewed in all dimensions concurrently. This makes some operations re. language simpler (e.g. finding translations is just a slice range), but the idea is that it should also be relatively inexpensive to add more dimensions if needed (e.g. role).
Fixes #10169
Fixes #10364
Fixes #10482
Fixes #10630
Fixes #10656
Fixes #10694
Fixes #10918
Fixes #11262
Fixes #11439
Fixes #11453
Fixes #11457
Fixes #11466
Fixes #11540
Fixes #11551
Fixes #11556
Fixes #11654
Fixes #11661
Fixes #11663
Fixes #11664
Fixes #11669
Fixes #11671
Fixes #11807
Fixes #11808
Fixes #11809
Fixes #11815
Fixes #11840
Fixes #11853
Fixes #11860
Fixes #11883
Fixes #11904
Fixes #7388
Fixes #7425
Fixes #7436
Fixes #7544
Fixes #7882
Fixes #7960
Fixes #8255
Fixes #8307
Fixes #8863
Fixes #8927
Fixes #9192
Fixes #9324
2023-12-24 13:11:05 -05:00
|
|
|
"math/rand"
|
2022-02-09 07:41:04 -05:00
|
|
|
"os"
|
|
|
|
"path/filepath"
|
2022-03-17 12:22:34 -04:00
|
|
|
"regexp"
|
2024-02-07 04:30:32 -05:00
|
|
|
"sort"
|
2022-02-09 07:41:04 -05:00
|
|
|
"strings"
|
|
|
|
"sync"
|
|
|
|
"testing"
|
|
|
|
|
2023-06-16 02:17:42 -04:00
|
|
|
"github.com/bep/logg"
|
2022-02-09 07:41:04 -05:00
|
|
|
|
|
|
|
qt "github.com/frankban/quicktest"
|
|
|
|
"github.com/fsnotify/fsnotify"
|
|
|
|
"github.com/gohugoio/hugo/common/herrors"
|
|
|
|
"github.com/gohugoio/hugo/common/hexec"
|
|
|
|
"github.com/gohugoio/hugo/common/loggers"
|
2023-01-04 12:24:36 -05:00
|
|
|
"github.com/gohugoio/hugo/common/maps"
|
2022-02-09 07:41:04 -05:00
|
|
|
"github.com/gohugoio/hugo/config"
|
2023-01-04 12:24:36 -05:00
|
|
|
"github.com/gohugoio/hugo/config/allconfig"
|
2022-02-09 07:41:04 -05:00
|
|
|
"github.com/gohugoio/hugo/config/security"
|
|
|
|
"github.com/gohugoio/hugo/deps"
|
|
|
|
"github.com/gohugoio/hugo/helpers"
|
|
|
|
"github.com/gohugoio/hugo/htesting"
|
|
|
|
"github.com/gohugoio/hugo/hugofs"
|
|
|
|
"github.com/spf13/afero"
|
all: Rework page store, add a dynacache, improve partial rebuilds, and some general spring cleaning
There are some breaking changes in this commit, see #11455.
Closes #11455
Closes #11549
This fixes a set of bugs (see issue list) and it is also paying some technical debt accumulated over the years. We now build with Staticcheck enabled in the CI build.
The performance should be about the same as before for regular sized Hugo sites, but it should perform and scale much better to larger data sets, as objects that uses lots of memory (e.g. rendered Markdown, big JSON files read into maps with transform.Unmarshal etc.) will now get automatically garbage collected if needed. Performance on partial rebuilds when running the server in fast render mode should be the same, but the change detection should be much more accurate.
A list of the notable new features:
* A new dependency tracker that covers (almost) all of Hugo's API and is used to do fine grained partial rebuilds when running the server.
* A new and simpler tree document store which allows fast lookups and prefix-walking in all dimensions (e.g. language) concurrently.
* You can now configure an upper memory limit allowing for much larger data sets and/or running on lower specced PCs.
We have lifted the "no resources in sub folders" restriction for branch bundles (e.g. sections).
Memory Limit
* Hugos will, by default, set aside a quarter of the total system memory, but you can set this via the OS environment variable HUGO_MEMORYLIMIT (in gigabytes). This is backed by a partitioned LRU cache used throughout Hugo. A cache that gets dynamically resized in low memory situations, allowing Go's Garbage Collector to free the memory.
New Dependency Tracker: Hugo has had a rule based coarse grained approach to server rebuilds that has worked mostly pretty well, but there have been some surprises (e.g. stale content). This is now revamped with a new dependency tracker that can quickly calculate the delta given a changed resource (e.g. a content file, template, JS file etc.). This handles transitive relations, e.g. $page -> js.Build -> JS import, or $page1.Content -> render hook -> site.GetPage -> $page2.Title, or $page1.Content -> shortcode -> partial -> site.RegularPages -> $page2.Content -> shortcode ..., and should also handle changes to aggregated values (e.g. site.Lastmod) effectively.
This covers all of Hugo's API with 2 known exceptions (a list that may not be fully exhaustive):
Changes to files loaded with template func os.ReadFile may not be handled correctly. We recommend loading resources with resources.Get
Changes to Hugo objects (e.g. Page) passed in the template context to lang.Translate may not be detected correctly. We recommend having simple i18n templates without too much data context passed in other than simple types such as strings and numbers.
Note that the cachebuster configuration (when A changes then rebuild B) works well with the above, but we recommend that you revise that configuration, as it in most situations should not be needed. One example where it is still needed is with TailwindCSS and using changes to hugo_stats.json to trigger new CSS rebuilds.
Document Store: Previously, a little simplified, we split the document store (where we store pages and resources) in a tree per language. This worked pretty well, but the structure made some operations harder than they needed to be. We have now restructured it into one Radix tree for all languages. Internally the language is considered to be a dimension of that tree, and the tree can be viewed in all dimensions concurrently. This makes some operations re. language simpler (e.g. finding translations is just a slice range), but the idea is that it should also be relatively inexpensive to add more dimensions if needed (e.g. role).
Fixes #10169
Fixes #10364
Fixes #10482
Fixes #10630
Fixes #10656
Fixes #10694
Fixes #10918
Fixes #11262
Fixes #11439
Fixes #11453
Fixes #11457
Fixes #11466
Fixes #11540
Fixes #11551
Fixes #11556
Fixes #11654
Fixes #11661
Fixes #11663
Fixes #11664
Fixes #11669
Fixes #11671
Fixes #11807
Fixes #11808
Fixes #11809
Fixes #11815
Fixes #11840
Fixes #11853
Fixes #11860
Fixes #11883
Fixes #11904
Fixes #7388
Fixes #7425
Fixes #7436
Fixes #7544
Fixes #7882
Fixes #7960
Fixes #8255
Fixes #8307
Fixes #8863
Fixes #8927
Fixes #9192
Fixes #9324
2023-12-24 13:11:05 -05:00
|
|
|
"golang.org/x/text/unicode/norm"
|
2022-02-09 07:41:04 -05:00
|
|
|
"golang.org/x/tools/txtar"
|
|
|
|
)
|
|
|
|
|
all: Rework page store, add a dynacache, improve partial rebuilds, and some general spring cleaning
There are some breaking changes in this commit, see #11455.
Closes #11455
Closes #11549
This fixes a set of bugs (see issue list) and it is also paying some technical debt accumulated over the years. We now build with Staticcheck enabled in the CI build.
The performance should be about the same as before for regular sized Hugo sites, but it should perform and scale much better to larger data sets, as objects that uses lots of memory (e.g. rendered Markdown, big JSON files read into maps with transform.Unmarshal etc.) will now get automatically garbage collected if needed. Performance on partial rebuilds when running the server in fast render mode should be the same, but the change detection should be much more accurate.
A list of the notable new features:
* A new dependency tracker that covers (almost) all of Hugo's API and is used to do fine grained partial rebuilds when running the server.
* A new and simpler tree document store which allows fast lookups and prefix-walking in all dimensions (e.g. language) concurrently.
* You can now configure an upper memory limit allowing for much larger data sets and/or running on lower specced PCs.
We have lifted the "no resources in sub folders" restriction for branch bundles (e.g. sections).
Memory Limit
* Hugos will, by default, set aside a quarter of the total system memory, but you can set this via the OS environment variable HUGO_MEMORYLIMIT (in gigabytes). This is backed by a partitioned LRU cache used throughout Hugo. A cache that gets dynamically resized in low memory situations, allowing Go's Garbage Collector to free the memory.
New Dependency Tracker: Hugo has had a rule based coarse grained approach to server rebuilds that has worked mostly pretty well, but there have been some surprises (e.g. stale content). This is now revamped with a new dependency tracker that can quickly calculate the delta given a changed resource (e.g. a content file, template, JS file etc.). This handles transitive relations, e.g. $page -> js.Build -> JS import, or $page1.Content -> render hook -> site.GetPage -> $page2.Title, or $page1.Content -> shortcode -> partial -> site.RegularPages -> $page2.Content -> shortcode ..., and should also handle changes to aggregated values (e.g. site.Lastmod) effectively.
This covers all of Hugo's API with 2 known exceptions (a list that may not be fully exhaustive):
Changes to files loaded with template func os.ReadFile may not be handled correctly. We recommend loading resources with resources.Get
Changes to Hugo objects (e.g. Page) passed in the template context to lang.Translate may not be detected correctly. We recommend having simple i18n templates without too much data context passed in other than simple types such as strings and numbers.
Note that the cachebuster configuration (when A changes then rebuild B) works well with the above, but we recommend that you revise that configuration, as it in most situations should not be needed. One example where it is still needed is with TailwindCSS and using changes to hugo_stats.json to trigger new CSS rebuilds.
Document Store: Previously, a little simplified, we split the document store (where we store pages and resources) in a tree per language. This worked pretty well, but the structure made some operations harder than they needed to be. We have now restructured it into one Radix tree for all languages. Internally the language is considered to be a dimension of that tree, and the tree can be viewed in all dimensions concurrently. This makes some operations re. language simpler (e.g. finding translations is just a slice range), but the idea is that it should also be relatively inexpensive to add more dimensions if needed (e.g. role).
Fixes #10169
Fixes #10364
Fixes #10482
Fixes #10630
Fixes #10656
Fixes #10694
Fixes #10918
Fixes #11262
Fixes #11439
Fixes #11453
Fixes #11457
Fixes #11466
Fixes #11540
Fixes #11551
Fixes #11556
Fixes #11654
Fixes #11661
Fixes #11663
Fixes #11664
Fixes #11669
Fixes #11671
Fixes #11807
Fixes #11808
Fixes #11809
Fixes #11815
Fixes #11840
Fixes #11853
Fixes #11860
Fixes #11883
Fixes #11904
Fixes #7388
Fixes #7425
Fixes #7436
Fixes #7544
Fixes #7882
Fixes #7960
Fixes #8255
Fixes #8307
Fixes #8863
Fixes #8927
Fixes #9192
Fixes #9324
2023-12-24 13:11:05 -05:00
|
|
|
type TestOpt func(*IntegrationTestConfig)
|
|
|
|
|
|
|
|
func TestOptRunning() TestOpt {
|
|
|
|
return func(c *IntegrationTestConfig) {
|
|
|
|
c.Running = true
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
// Enable tracing in integration tests.
|
|
|
|
// THis should only be used during development and not committed to the repo.
|
|
|
|
func TestOptTrace() TestOpt {
|
|
|
|
return func(c *IntegrationTestConfig) {
|
|
|
|
c.LogLevel = logg.LevelTrace
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
// TestOptDebug will enable debug logging in integration tests.
|
|
|
|
func TestOptDebug() TestOpt {
|
|
|
|
return func(c *IntegrationTestConfig) {
|
|
|
|
c.LogLevel = logg.LevelDebug
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
2024-01-30 03:23:21 -05:00
|
|
|
// TestOptWarn will enable warn logging in integration tests.
|
|
|
|
func TestOptWarn() TestOpt {
|
|
|
|
return func(c *IntegrationTestConfig) {
|
|
|
|
c.LogLevel = logg.LevelWarn
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
all: Rework page store, add a dynacache, improve partial rebuilds, and some general spring cleaning
There are some breaking changes in this commit, see #11455.
Closes #11455
Closes #11549
This fixes a set of bugs (see issue list) and it is also paying some technical debt accumulated over the years. We now build with Staticcheck enabled in the CI build.
The performance should be about the same as before for regular sized Hugo sites, but it should perform and scale much better to larger data sets, as objects that uses lots of memory (e.g. rendered Markdown, big JSON files read into maps with transform.Unmarshal etc.) will now get automatically garbage collected if needed. Performance on partial rebuilds when running the server in fast render mode should be the same, but the change detection should be much more accurate.
A list of the notable new features:
* A new dependency tracker that covers (almost) all of Hugo's API and is used to do fine grained partial rebuilds when running the server.
* A new and simpler tree document store which allows fast lookups and prefix-walking in all dimensions (e.g. language) concurrently.
* You can now configure an upper memory limit allowing for much larger data sets and/or running on lower specced PCs.
We have lifted the "no resources in sub folders" restriction for branch bundles (e.g. sections).
Memory Limit
* Hugos will, by default, set aside a quarter of the total system memory, but you can set this via the OS environment variable HUGO_MEMORYLIMIT (in gigabytes). This is backed by a partitioned LRU cache used throughout Hugo. A cache that gets dynamically resized in low memory situations, allowing Go's Garbage Collector to free the memory.
New Dependency Tracker: Hugo has had a rule based coarse grained approach to server rebuilds that has worked mostly pretty well, but there have been some surprises (e.g. stale content). This is now revamped with a new dependency tracker that can quickly calculate the delta given a changed resource (e.g. a content file, template, JS file etc.). This handles transitive relations, e.g. $page -> js.Build -> JS import, or $page1.Content -> render hook -> site.GetPage -> $page2.Title, or $page1.Content -> shortcode -> partial -> site.RegularPages -> $page2.Content -> shortcode ..., and should also handle changes to aggregated values (e.g. site.Lastmod) effectively.
This covers all of Hugo's API with 2 known exceptions (a list that may not be fully exhaustive):
Changes to files loaded with template func os.ReadFile may not be handled correctly. We recommend loading resources with resources.Get
Changes to Hugo objects (e.g. Page) passed in the template context to lang.Translate may not be detected correctly. We recommend having simple i18n templates without too much data context passed in other than simple types such as strings and numbers.
Note that the cachebuster configuration (when A changes then rebuild B) works well with the above, but we recommend that you revise that configuration, as it in most situations should not be needed. One example where it is still needed is with TailwindCSS and using changes to hugo_stats.json to trigger new CSS rebuilds.
Document Store: Previously, a little simplified, we split the document store (where we store pages and resources) in a tree per language. This worked pretty well, but the structure made some operations harder than they needed to be. We have now restructured it into one Radix tree for all languages. Internally the language is considered to be a dimension of that tree, and the tree can be viewed in all dimensions concurrently. This makes some operations re. language simpler (e.g. finding translations is just a slice range), but the idea is that it should also be relatively inexpensive to add more dimensions if needed (e.g. role).
Fixes #10169
Fixes #10364
Fixes #10482
Fixes #10630
Fixes #10656
Fixes #10694
Fixes #10918
Fixes #11262
Fixes #11439
Fixes #11453
Fixes #11457
Fixes #11466
Fixes #11540
Fixes #11551
Fixes #11556
Fixes #11654
Fixes #11661
Fixes #11663
Fixes #11664
Fixes #11669
Fixes #11671
Fixes #11807
Fixes #11808
Fixes #11809
Fixes #11815
Fixes #11840
Fixes #11853
Fixes #11860
Fixes #11883
Fixes #11904
Fixes #7388
Fixes #7425
Fixes #7436
Fixes #7544
Fixes #7882
Fixes #7960
Fixes #8255
Fixes #8307
Fixes #8863
Fixes #8927
Fixes #9192
Fixes #9324
2023-12-24 13:11:05 -05:00
|
|
|
// TestOptWithNFDOnDarwin will normalize the Unicode filenames to NFD on Darwin.
|
|
|
|
func TestOptWithNFDOnDarwin() TestOpt {
|
|
|
|
return func(c *IntegrationTestConfig) {
|
|
|
|
c.NFDFormOnDarwin = true
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
// TestOptWithWorkingDir allows setting any config optiona as a function al option.
|
|
|
|
func TestOptWithConfig(fn func(c *IntegrationTestConfig)) TestOpt {
|
|
|
|
return func(c *IntegrationTestConfig) {
|
|
|
|
fn(c)
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
2023-10-18 12:54:15 -04:00
|
|
|
// Test is a convenience method to create a new IntegrationTestBuilder from some files and run a build.
|
all: Rework page store, add a dynacache, improve partial rebuilds, and some general spring cleaning
There are some breaking changes in this commit, see #11455.
Closes #11455
Closes #11549
This fixes a set of bugs (see issue list) and it is also paying some technical debt accumulated over the years. We now build with Staticcheck enabled in the CI build.
The performance should be about the same as before for regular sized Hugo sites, but it should perform and scale much better to larger data sets, as objects that uses lots of memory (e.g. rendered Markdown, big JSON files read into maps with transform.Unmarshal etc.) will now get automatically garbage collected if needed. Performance on partial rebuilds when running the server in fast render mode should be the same, but the change detection should be much more accurate.
A list of the notable new features:
* A new dependency tracker that covers (almost) all of Hugo's API and is used to do fine grained partial rebuilds when running the server.
* A new and simpler tree document store which allows fast lookups and prefix-walking in all dimensions (e.g. language) concurrently.
* You can now configure an upper memory limit allowing for much larger data sets and/or running on lower specced PCs.
We have lifted the "no resources in sub folders" restriction for branch bundles (e.g. sections).
Memory Limit
* Hugos will, by default, set aside a quarter of the total system memory, but you can set this via the OS environment variable HUGO_MEMORYLIMIT (in gigabytes). This is backed by a partitioned LRU cache used throughout Hugo. A cache that gets dynamically resized in low memory situations, allowing Go's Garbage Collector to free the memory.
New Dependency Tracker: Hugo has had a rule based coarse grained approach to server rebuilds that has worked mostly pretty well, but there have been some surprises (e.g. stale content). This is now revamped with a new dependency tracker that can quickly calculate the delta given a changed resource (e.g. a content file, template, JS file etc.). This handles transitive relations, e.g. $page -> js.Build -> JS import, or $page1.Content -> render hook -> site.GetPage -> $page2.Title, or $page1.Content -> shortcode -> partial -> site.RegularPages -> $page2.Content -> shortcode ..., and should also handle changes to aggregated values (e.g. site.Lastmod) effectively.
This covers all of Hugo's API with 2 known exceptions (a list that may not be fully exhaustive):
Changes to files loaded with template func os.ReadFile may not be handled correctly. We recommend loading resources with resources.Get
Changes to Hugo objects (e.g. Page) passed in the template context to lang.Translate may not be detected correctly. We recommend having simple i18n templates without too much data context passed in other than simple types such as strings and numbers.
Note that the cachebuster configuration (when A changes then rebuild B) works well with the above, but we recommend that you revise that configuration, as it in most situations should not be needed. One example where it is still needed is with TailwindCSS and using changes to hugo_stats.json to trigger new CSS rebuilds.
Document Store: Previously, a little simplified, we split the document store (where we store pages and resources) in a tree per language. This worked pretty well, but the structure made some operations harder than they needed to be. We have now restructured it into one Radix tree for all languages. Internally the language is considered to be a dimension of that tree, and the tree can be viewed in all dimensions concurrently. This makes some operations re. language simpler (e.g. finding translations is just a slice range), but the idea is that it should also be relatively inexpensive to add more dimensions if needed (e.g. role).
Fixes #10169
Fixes #10364
Fixes #10482
Fixes #10630
Fixes #10656
Fixes #10694
Fixes #10918
Fixes #11262
Fixes #11439
Fixes #11453
Fixes #11457
Fixes #11466
Fixes #11540
Fixes #11551
Fixes #11556
Fixes #11654
Fixes #11661
Fixes #11663
Fixes #11664
Fixes #11669
Fixes #11671
Fixes #11807
Fixes #11808
Fixes #11809
Fixes #11815
Fixes #11840
Fixes #11853
Fixes #11860
Fixes #11883
Fixes #11904
Fixes #7388
Fixes #7425
Fixes #7436
Fixes #7544
Fixes #7882
Fixes #7960
Fixes #8255
Fixes #8307
Fixes #8863
Fixes #8927
Fixes #9192
Fixes #9324
2023-12-24 13:11:05 -05:00
|
|
|
func Test(t testing.TB, files string, opts ...TestOpt) *IntegrationTestBuilder {
|
|
|
|
cfg := IntegrationTestConfig{T: t, TxtarString: files}
|
|
|
|
for _, o := range opts {
|
|
|
|
o(&cfg)
|
|
|
|
}
|
|
|
|
return NewIntegrationTestBuilder(cfg).Build()
|
2023-10-18 12:54:15 -04:00
|
|
|
}
|
|
|
|
|
2024-01-29 04:02:24 -05:00
|
|
|
// TestE is the same as Test, but returns an error instead of failing the test.
|
|
|
|
func TestE(t testing.TB, files string, opts ...TestOpt) (*IntegrationTestBuilder, error) {
|
|
|
|
cfg := IntegrationTestConfig{T: t, TxtarString: files}
|
|
|
|
for _, o := range opts {
|
|
|
|
o(&cfg)
|
|
|
|
}
|
|
|
|
return NewIntegrationTestBuilder(cfg).BuildE()
|
|
|
|
}
|
|
|
|
|
2023-10-18 12:54:15 -04:00
|
|
|
// TestRunning is a convenience method to create a new IntegrationTestBuilder from some files with Running set to true and run a build.
|
all: Rework page store, add a dynacache, improve partial rebuilds, and some general spring cleaning
There are some breaking changes in this commit, see #11455.
Closes #11455
Closes #11549
This fixes a set of bugs (see issue list) and it is also paying some technical debt accumulated over the years. We now build with Staticcheck enabled in the CI build.
The performance should be about the same as before for regular sized Hugo sites, but it should perform and scale much better to larger data sets, as objects that uses lots of memory (e.g. rendered Markdown, big JSON files read into maps with transform.Unmarshal etc.) will now get automatically garbage collected if needed. Performance on partial rebuilds when running the server in fast render mode should be the same, but the change detection should be much more accurate.
A list of the notable new features:
* A new dependency tracker that covers (almost) all of Hugo's API and is used to do fine grained partial rebuilds when running the server.
* A new and simpler tree document store which allows fast lookups and prefix-walking in all dimensions (e.g. language) concurrently.
* You can now configure an upper memory limit allowing for much larger data sets and/or running on lower specced PCs.
We have lifted the "no resources in sub folders" restriction for branch bundles (e.g. sections).
Memory Limit
* Hugos will, by default, set aside a quarter of the total system memory, but you can set this via the OS environment variable HUGO_MEMORYLIMIT (in gigabytes). This is backed by a partitioned LRU cache used throughout Hugo. A cache that gets dynamically resized in low memory situations, allowing Go's Garbage Collector to free the memory.
New Dependency Tracker: Hugo has had a rule based coarse grained approach to server rebuilds that has worked mostly pretty well, but there have been some surprises (e.g. stale content). This is now revamped with a new dependency tracker that can quickly calculate the delta given a changed resource (e.g. a content file, template, JS file etc.). This handles transitive relations, e.g. $page -> js.Build -> JS import, or $page1.Content -> render hook -> site.GetPage -> $page2.Title, or $page1.Content -> shortcode -> partial -> site.RegularPages -> $page2.Content -> shortcode ..., and should also handle changes to aggregated values (e.g. site.Lastmod) effectively.
This covers all of Hugo's API with 2 known exceptions (a list that may not be fully exhaustive):
Changes to files loaded with template func os.ReadFile may not be handled correctly. We recommend loading resources with resources.Get
Changes to Hugo objects (e.g. Page) passed in the template context to lang.Translate may not be detected correctly. We recommend having simple i18n templates without too much data context passed in other than simple types such as strings and numbers.
Note that the cachebuster configuration (when A changes then rebuild B) works well with the above, but we recommend that you revise that configuration, as it in most situations should not be needed. One example where it is still needed is with TailwindCSS and using changes to hugo_stats.json to trigger new CSS rebuilds.
Document Store: Previously, a little simplified, we split the document store (where we store pages and resources) in a tree per language. This worked pretty well, but the structure made some operations harder than they needed to be. We have now restructured it into one Radix tree for all languages. Internally the language is considered to be a dimension of that tree, and the tree can be viewed in all dimensions concurrently. This makes some operations re. language simpler (e.g. finding translations is just a slice range), but the idea is that it should also be relatively inexpensive to add more dimensions if needed (e.g. role).
Fixes #10169
Fixes #10364
Fixes #10482
Fixes #10630
Fixes #10656
Fixes #10694
Fixes #10918
Fixes #11262
Fixes #11439
Fixes #11453
Fixes #11457
Fixes #11466
Fixes #11540
Fixes #11551
Fixes #11556
Fixes #11654
Fixes #11661
Fixes #11663
Fixes #11664
Fixes #11669
Fixes #11671
Fixes #11807
Fixes #11808
Fixes #11809
Fixes #11815
Fixes #11840
Fixes #11853
Fixes #11860
Fixes #11883
Fixes #11904
Fixes #7388
Fixes #7425
Fixes #7436
Fixes #7544
Fixes #7882
Fixes #7960
Fixes #8255
Fixes #8307
Fixes #8863
Fixes #8927
Fixes #9192
Fixes #9324
2023-12-24 13:11:05 -05:00
|
|
|
// Deprecated: Use Test with TestOptRunning instead.
|
|
|
|
func TestRunning(t testing.TB, files string, opts ...TestOpt) *IntegrationTestBuilder {
|
|
|
|
cfg := IntegrationTestConfig{T: t, TxtarString: files, Running: true}
|
|
|
|
for _, o := range opts {
|
|
|
|
o(&cfg)
|
|
|
|
}
|
|
|
|
return NewIntegrationTestBuilder(cfg).Build()
|
2023-10-18 12:54:15 -04:00
|
|
|
}
|
|
|
|
|
2024-01-28 05:41:59 -05:00
|
|
|
// In most cases you should not use this function directly, but the Test or TestRunning function.
|
2022-02-09 07:41:04 -05:00
|
|
|
func NewIntegrationTestBuilder(conf IntegrationTestConfig) *IntegrationTestBuilder {
|
2022-02-26 04:42:21 -05:00
|
|
|
// Code fences.
|
|
|
|
conf.TxtarString = strings.ReplaceAll(conf.TxtarString, "§§§", "```")
|
2023-02-24 01:23:10 -05:00
|
|
|
// Multiline strings.
|
|
|
|
conf.TxtarString = strings.ReplaceAll(conf.TxtarString, "§§", "`")
|
2022-02-26 04:42:21 -05:00
|
|
|
|
2022-02-09 07:41:04 -05:00
|
|
|
data := txtar.Parse([]byte(conf.TxtarString))
|
|
|
|
|
all: Rework page store, add a dynacache, improve partial rebuilds, and some general spring cleaning
There are some breaking changes in this commit, see #11455.
Closes #11455
Closes #11549
This fixes a set of bugs (see issue list) and it is also paying some technical debt accumulated over the years. We now build with Staticcheck enabled in the CI build.
The performance should be about the same as before for regular sized Hugo sites, but it should perform and scale much better to larger data sets, as objects that uses lots of memory (e.g. rendered Markdown, big JSON files read into maps with transform.Unmarshal etc.) will now get automatically garbage collected if needed. Performance on partial rebuilds when running the server in fast render mode should be the same, but the change detection should be much more accurate.
A list of the notable new features:
* A new dependency tracker that covers (almost) all of Hugo's API and is used to do fine grained partial rebuilds when running the server.
* A new and simpler tree document store which allows fast lookups and prefix-walking in all dimensions (e.g. language) concurrently.
* You can now configure an upper memory limit allowing for much larger data sets and/or running on lower specced PCs.
We have lifted the "no resources in sub folders" restriction for branch bundles (e.g. sections).
Memory Limit
* Hugos will, by default, set aside a quarter of the total system memory, but you can set this via the OS environment variable HUGO_MEMORYLIMIT (in gigabytes). This is backed by a partitioned LRU cache used throughout Hugo. A cache that gets dynamically resized in low memory situations, allowing Go's Garbage Collector to free the memory.
New Dependency Tracker: Hugo has had a rule based coarse grained approach to server rebuilds that has worked mostly pretty well, but there have been some surprises (e.g. stale content). This is now revamped with a new dependency tracker that can quickly calculate the delta given a changed resource (e.g. a content file, template, JS file etc.). This handles transitive relations, e.g. $page -> js.Build -> JS import, or $page1.Content -> render hook -> site.GetPage -> $page2.Title, or $page1.Content -> shortcode -> partial -> site.RegularPages -> $page2.Content -> shortcode ..., and should also handle changes to aggregated values (e.g. site.Lastmod) effectively.
This covers all of Hugo's API with 2 known exceptions (a list that may not be fully exhaustive):
Changes to files loaded with template func os.ReadFile may not be handled correctly. We recommend loading resources with resources.Get
Changes to Hugo objects (e.g. Page) passed in the template context to lang.Translate may not be detected correctly. We recommend having simple i18n templates without too much data context passed in other than simple types such as strings and numbers.
Note that the cachebuster configuration (when A changes then rebuild B) works well with the above, but we recommend that you revise that configuration, as it in most situations should not be needed. One example where it is still needed is with TailwindCSS and using changes to hugo_stats.json to trigger new CSS rebuilds.
Document Store: Previously, a little simplified, we split the document store (where we store pages and resources) in a tree per language. This worked pretty well, but the structure made some operations harder than they needed to be. We have now restructured it into one Radix tree for all languages. Internally the language is considered to be a dimension of that tree, and the tree can be viewed in all dimensions concurrently. This makes some operations re. language simpler (e.g. finding translations is just a slice range), but the idea is that it should also be relatively inexpensive to add more dimensions if needed (e.g. role).
Fixes #10169
Fixes #10364
Fixes #10482
Fixes #10630
Fixes #10656
Fixes #10694
Fixes #10918
Fixes #11262
Fixes #11439
Fixes #11453
Fixes #11457
Fixes #11466
Fixes #11540
Fixes #11551
Fixes #11556
Fixes #11654
Fixes #11661
Fixes #11663
Fixes #11664
Fixes #11669
Fixes #11671
Fixes #11807
Fixes #11808
Fixes #11809
Fixes #11815
Fixes #11840
Fixes #11853
Fixes #11860
Fixes #11883
Fixes #11904
Fixes #7388
Fixes #7425
Fixes #7436
Fixes #7544
Fixes #7882
Fixes #7960
Fixes #8255
Fixes #8307
Fixes #8863
Fixes #8927
Fixes #9192
Fixes #9324
2023-12-24 13:11:05 -05:00
|
|
|
if conf.NFDFormOnDarwin {
|
|
|
|
for i, f := range data.Files {
|
|
|
|
data.Files[i].Name = norm.NFD.String(f.Name)
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
2022-02-09 07:41:04 -05:00
|
|
|
c, ok := conf.T.(*qt.C)
|
|
|
|
if !ok {
|
|
|
|
c = qt.New(conf.T)
|
|
|
|
}
|
|
|
|
|
|
|
|
if conf.NeedsOsFS {
|
2022-05-27 07:23:37 -04:00
|
|
|
if !filepath.IsAbs(conf.WorkingDir) {
|
|
|
|
tempDir, clean, err := htesting.CreateTempDir(hugofs.Os, "hugo-integration-test")
|
|
|
|
c.Assert(err, qt.IsNil)
|
|
|
|
conf.WorkingDir = filepath.Join(tempDir, conf.WorkingDir)
|
|
|
|
if !conf.PrintAndKeepTempDir {
|
|
|
|
c.Cleanup(clean)
|
|
|
|
} else {
|
|
|
|
fmt.Println("\nUsing WorkingDir dir:", conf.WorkingDir)
|
|
|
|
}
|
2022-02-09 07:41:04 -05:00
|
|
|
}
|
2022-03-21 04:35:15 -04:00
|
|
|
} else if conf.WorkingDir == "" {
|
|
|
|
conf.WorkingDir = helpers.FilePathSeparator
|
2022-02-09 07:41:04 -05:00
|
|
|
}
|
|
|
|
|
|
|
|
return &IntegrationTestBuilder{
|
|
|
|
Cfg: conf,
|
|
|
|
C: c,
|
|
|
|
data: data,
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
// IntegrationTestBuilder is a (partial) rewrite of sitesBuilder.
|
|
|
|
// The main problem with the "old" one was that it was that the test data was often a little hidden,
|
|
|
|
// so it became hard to look at a test and determine what it should do, especially coming back to the
|
|
|
|
// test after a year or so.
|
|
|
|
type IntegrationTestBuilder struct {
|
|
|
|
*qt.C
|
|
|
|
|
|
|
|
data *txtar.Archive
|
|
|
|
|
|
|
|
fs *hugofs.Fs
|
|
|
|
H *HugoSites
|
|
|
|
|
|
|
|
Cfg IntegrationTestConfig
|
|
|
|
|
|
|
|
changedFiles []string
|
|
|
|
createdFiles []string
|
|
|
|
removedFiles []string
|
|
|
|
renamedFiles []string
|
all: Rework page store, add a dynacache, improve partial rebuilds, and some general spring cleaning
There are some breaking changes in this commit, see #11455.
Closes #11455
Closes #11549
This fixes a set of bugs (see issue list) and it is also paying some technical debt accumulated over the years. We now build with Staticcheck enabled in the CI build.
The performance should be about the same as before for regular sized Hugo sites, but it should perform and scale much better to larger data sets, as objects that uses lots of memory (e.g. rendered Markdown, big JSON files read into maps with transform.Unmarshal etc.) will now get automatically garbage collected if needed. Performance on partial rebuilds when running the server in fast render mode should be the same, but the change detection should be much more accurate.
A list of the notable new features:
* A new dependency tracker that covers (almost) all of Hugo's API and is used to do fine grained partial rebuilds when running the server.
* A new and simpler tree document store which allows fast lookups and prefix-walking in all dimensions (e.g. language) concurrently.
* You can now configure an upper memory limit allowing for much larger data sets and/or running on lower specced PCs.
We have lifted the "no resources in sub folders" restriction for branch bundles (e.g. sections).
Memory Limit
* Hugos will, by default, set aside a quarter of the total system memory, but you can set this via the OS environment variable HUGO_MEMORYLIMIT (in gigabytes). This is backed by a partitioned LRU cache used throughout Hugo. A cache that gets dynamically resized in low memory situations, allowing Go's Garbage Collector to free the memory.
New Dependency Tracker: Hugo has had a rule based coarse grained approach to server rebuilds that has worked mostly pretty well, but there have been some surprises (e.g. stale content). This is now revamped with a new dependency tracker that can quickly calculate the delta given a changed resource (e.g. a content file, template, JS file etc.). This handles transitive relations, e.g. $page -> js.Build -> JS import, or $page1.Content -> render hook -> site.GetPage -> $page2.Title, or $page1.Content -> shortcode -> partial -> site.RegularPages -> $page2.Content -> shortcode ..., and should also handle changes to aggregated values (e.g. site.Lastmod) effectively.
This covers all of Hugo's API with 2 known exceptions (a list that may not be fully exhaustive):
Changes to files loaded with template func os.ReadFile may not be handled correctly. We recommend loading resources with resources.Get
Changes to Hugo objects (e.g. Page) passed in the template context to lang.Translate may not be detected correctly. We recommend having simple i18n templates without too much data context passed in other than simple types such as strings and numbers.
Note that the cachebuster configuration (when A changes then rebuild B) works well with the above, but we recommend that you revise that configuration, as it in most situations should not be needed. One example where it is still needed is with TailwindCSS and using changes to hugo_stats.json to trigger new CSS rebuilds.
Document Store: Previously, a little simplified, we split the document store (where we store pages and resources) in a tree per language. This worked pretty well, but the structure made some operations harder than they needed to be. We have now restructured it into one Radix tree for all languages. Internally the language is considered to be a dimension of that tree, and the tree can be viewed in all dimensions concurrently. This makes some operations re. language simpler (e.g. finding translations is just a slice range), but the idea is that it should also be relatively inexpensive to add more dimensions if needed (e.g. role).
Fixes #10169
Fixes #10364
Fixes #10482
Fixes #10630
Fixes #10656
Fixes #10694
Fixes #10918
Fixes #11262
Fixes #11439
Fixes #11453
Fixes #11457
Fixes #11466
Fixes #11540
Fixes #11551
Fixes #11556
Fixes #11654
Fixes #11661
Fixes #11663
Fixes #11664
Fixes #11669
Fixes #11671
Fixes #11807
Fixes #11808
Fixes #11809
Fixes #11815
Fixes #11840
Fixes #11853
Fixes #11860
Fixes #11883
Fixes #11904
Fixes #7388
Fixes #7425
Fixes #7436
Fixes #7544
Fixes #7882
Fixes #7960
Fixes #8255
Fixes #8307
Fixes #8863
Fixes #8927
Fixes #9192
Fixes #9324
2023-12-24 13:11:05 -05:00
|
|
|
renamedDirs []string
|
2022-02-09 07:41:04 -05:00
|
|
|
|
|
|
|
buildCount int
|
2023-03-04 05:09:03 -05:00
|
|
|
GCCount int
|
all: Rework page store, add a dynacache, improve partial rebuilds, and some general spring cleaning
There are some breaking changes in this commit, see #11455.
Closes #11455
Closes #11549
This fixes a set of bugs (see issue list) and it is also paying some technical debt accumulated over the years. We now build with Staticcheck enabled in the CI build.
The performance should be about the same as before for regular sized Hugo sites, but it should perform and scale much better to larger data sets, as objects that uses lots of memory (e.g. rendered Markdown, big JSON files read into maps with transform.Unmarshal etc.) will now get automatically garbage collected if needed. Performance on partial rebuilds when running the server in fast render mode should be the same, but the change detection should be much more accurate.
A list of the notable new features:
* A new dependency tracker that covers (almost) all of Hugo's API and is used to do fine grained partial rebuilds when running the server.
* A new and simpler tree document store which allows fast lookups and prefix-walking in all dimensions (e.g. language) concurrently.
* You can now configure an upper memory limit allowing for much larger data sets and/or running on lower specced PCs.
We have lifted the "no resources in sub folders" restriction for branch bundles (e.g. sections).
Memory Limit
* Hugos will, by default, set aside a quarter of the total system memory, but you can set this via the OS environment variable HUGO_MEMORYLIMIT (in gigabytes). This is backed by a partitioned LRU cache used throughout Hugo. A cache that gets dynamically resized in low memory situations, allowing Go's Garbage Collector to free the memory.
New Dependency Tracker: Hugo has had a rule based coarse grained approach to server rebuilds that has worked mostly pretty well, but there have been some surprises (e.g. stale content). This is now revamped with a new dependency tracker that can quickly calculate the delta given a changed resource (e.g. a content file, template, JS file etc.). This handles transitive relations, e.g. $page -> js.Build -> JS import, or $page1.Content -> render hook -> site.GetPage -> $page2.Title, or $page1.Content -> shortcode -> partial -> site.RegularPages -> $page2.Content -> shortcode ..., and should also handle changes to aggregated values (e.g. site.Lastmod) effectively.
This covers all of Hugo's API with 2 known exceptions (a list that may not be fully exhaustive):
Changes to files loaded with template func os.ReadFile may not be handled correctly. We recommend loading resources with resources.Get
Changes to Hugo objects (e.g. Page) passed in the template context to lang.Translate may not be detected correctly. We recommend having simple i18n templates without too much data context passed in other than simple types such as strings and numbers.
Note that the cachebuster configuration (when A changes then rebuild B) works well with the above, but we recommend that you revise that configuration, as it in most situations should not be needed. One example where it is still needed is with TailwindCSS and using changes to hugo_stats.json to trigger new CSS rebuilds.
Document Store: Previously, a little simplified, we split the document store (where we store pages and resources) in a tree per language. This worked pretty well, but the structure made some operations harder than they needed to be. We have now restructured it into one Radix tree for all languages. Internally the language is considered to be a dimension of that tree, and the tree can be viewed in all dimensions concurrently. This makes some operations re. language simpler (e.g. finding translations is just a slice range), but the idea is that it should also be relatively inexpensive to add more dimensions if needed (e.g. role).
Fixes #10169
Fixes #10364
Fixes #10482
Fixes #10630
Fixes #10656
Fixes #10694
Fixes #10918
Fixes #11262
Fixes #11439
Fixes #11453
Fixes #11457
Fixes #11466
Fixes #11540
Fixes #11551
Fixes #11556
Fixes #11654
Fixes #11661
Fixes #11663
Fixes #11664
Fixes #11669
Fixes #11671
Fixes #11807
Fixes #11808
Fixes #11809
Fixes #11815
Fixes #11840
Fixes #11853
Fixes #11860
Fixes #11883
Fixes #11904
Fixes #7388
Fixes #7425
Fixes #7436
Fixes #7544
Fixes #7882
Fixes #7960
Fixes #8255
Fixes #8307
Fixes #8863
Fixes #8927
Fixes #9192
Fixes #9324
2023-12-24 13:11:05 -05:00
|
|
|
counters *buildCounters
|
2022-02-09 07:41:04 -05:00
|
|
|
logBuff lockingBuffer
|
|
|
|
|
|
|
|
builderInit sync.Once
|
|
|
|
}
|
|
|
|
|
|
|
|
type lockingBuffer struct {
|
|
|
|
sync.Mutex
|
|
|
|
bytes.Buffer
|
|
|
|
}
|
|
|
|
|
|
|
|
func (b *lockingBuffer) Write(p []byte) (n int, err error) {
|
|
|
|
b.Lock()
|
|
|
|
n, err = b.Buffer.Write(p)
|
|
|
|
b.Unlock()
|
|
|
|
return
|
|
|
|
}
|
|
|
|
|
2024-01-30 03:23:21 -05:00
|
|
|
func (s *IntegrationTestBuilder) AssertLogContains(els ...string) {
|
2022-02-09 07:41:04 -05:00
|
|
|
s.Helper()
|
2024-01-30 03:23:21 -05:00
|
|
|
for _, el := range els {
|
|
|
|
s.Assert(s.logBuff.String(), qt.Contains, el)
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
func (s *IntegrationTestBuilder) AssertLogNotContains(els ...string) {
|
|
|
|
s.Helper()
|
|
|
|
for _, el := range els {
|
|
|
|
s.Assert(s.logBuff.String(), qt.Not(qt.Contains), el)
|
|
|
|
}
|
2022-02-09 07:41:04 -05:00
|
|
|
}
|
|
|
|
|
2022-03-17 12:22:34 -04:00
|
|
|
func (s *IntegrationTestBuilder) AssertLogMatches(expression string) {
|
|
|
|
s.Helper()
|
|
|
|
re := regexp.MustCompile(expression)
|
|
|
|
s.Assert(re.MatchString(s.logBuff.String()), qt.IsTrue, qt.Commentf(s.logBuff.String()))
|
|
|
|
}
|
|
|
|
|
2022-02-09 07:41:04 -05:00
|
|
|
func (s *IntegrationTestBuilder) AssertBuildCountData(count int) {
|
|
|
|
s.Helper()
|
|
|
|
s.Assert(s.H.init.data.InitCount(), qt.Equals, count)
|
|
|
|
}
|
|
|
|
|
|
|
|
func (s *IntegrationTestBuilder) AssertBuildCountGitInfo(count int) {
|
|
|
|
s.Helper()
|
|
|
|
s.Assert(s.H.init.gitInfo.InitCount(), qt.Equals, count)
|
|
|
|
}
|
|
|
|
|
|
|
|
func (s *IntegrationTestBuilder) AssertBuildCountLayouts(count int) {
|
|
|
|
s.Helper()
|
|
|
|
s.Assert(s.H.init.layouts.InitCount(), qt.Equals, count)
|
|
|
|
}
|
|
|
|
|
2023-07-30 05:26:29 -04:00
|
|
|
func (s *IntegrationTestBuilder) AssertFileCount(dirname string, expected int) {
|
|
|
|
s.Helper()
|
|
|
|
fs := s.fs.WorkingDirReadOnly
|
|
|
|
count := 0
|
|
|
|
afero.Walk(fs, dirname, func(path string, info os.FileInfo, err error) error {
|
|
|
|
if err != nil {
|
|
|
|
return err
|
|
|
|
}
|
|
|
|
if info.IsDir() {
|
|
|
|
return nil
|
|
|
|
}
|
|
|
|
count++
|
|
|
|
return nil
|
|
|
|
})
|
|
|
|
s.Assert(count, qt.Equals, expected)
|
|
|
|
}
|
|
|
|
|
2022-02-09 07:41:04 -05:00
|
|
|
func (s *IntegrationTestBuilder) AssertFileContent(filename string, matches ...string) {
|
|
|
|
s.Helper()
|
|
|
|
content := strings.TrimSpace(s.FileContent(filename))
|
|
|
|
for _, m := range matches {
|
all: Rework page store, add a dynacache, improve partial rebuilds, and some general spring cleaning
There are some breaking changes in this commit, see #11455.
Closes #11455
Closes #11549
This fixes a set of bugs (see issue list) and it is also paying some technical debt accumulated over the years. We now build with Staticcheck enabled in the CI build.
The performance should be about the same as before for regular sized Hugo sites, but it should perform and scale much better to larger data sets, as objects that uses lots of memory (e.g. rendered Markdown, big JSON files read into maps with transform.Unmarshal etc.) will now get automatically garbage collected if needed. Performance on partial rebuilds when running the server in fast render mode should be the same, but the change detection should be much more accurate.
A list of the notable new features:
* A new dependency tracker that covers (almost) all of Hugo's API and is used to do fine grained partial rebuilds when running the server.
* A new and simpler tree document store which allows fast lookups and prefix-walking in all dimensions (e.g. language) concurrently.
* You can now configure an upper memory limit allowing for much larger data sets and/or running on lower specced PCs.
We have lifted the "no resources in sub folders" restriction for branch bundles (e.g. sections).
Memory Limit
* Hugos will, by default, set aside a quarter of the total system memory, but you can set this via the OS environment variable HUGO_MEMORYLIMIT (in gigabytes). This is backed by a partitioned LRU cache used throughout Hugo. A cache that gets dynamically resized in low memory situations, allowing Go's Garbage Collector to free the memory.
New Dependency Tracker: Hugo has had a rule based coarse grained approach to server rebuilds that has worked mostly pretty well, but there have been some surprises (e.g. stale content). This is now revamped with a new dependency tracker that can quickly calculate the delta given a changed resource (e.g. a content file, template, JS file etc.). This handles transitive relations, e.g. $page -> js.Build -> JS import, or $page1.Content -> render hook -> site.GetPage -> $page2.Title, or $page1.Content -> shortcode -> partial -> site.RegularPages -> $page2.Content -> shortcode ..., and should also handle changes to aggregated values (e.g. site.Lastmod) effectively.
This covers all of Hugo's API with 2 known exceptions (a list that may not be fully exhaustive):
Changes to files loaded with template func os.ReadFile may not be handled correctly. We recommend loading resources with resources.Get
Changes to Hugo objects (e.g. Page) passed in the template context to lang.Translate may not be detected correctly. We recommend having simple i18n templates without too much data context passed in other than simple types such as strings and numbers.
Note that the cachebuster configuration (when A changes then rebuild B) works well with the above, but we recommend that you revise that configuration, as it in most situations should not be needed. One example where it is still needed is with TailwindCSS and using changes to hugo_stats.json to trigger new CSS rebuilds.
Document Store: Previously, a little simplified, we split the document store (where we store pages and resources) in a tree per language. This worked pretty well, but the structure made some operations harder than they needed to be. We have now restructured it into one Radix tree for all languages. Internally the language is considered to be a dimension of that tree, and the tree can be viewed in all dimensions concurrently. This makes some operations re. language simpler (e.g. finding translations is just a slice range), but the idea is that it should also be relatively inexpensive to add more dimensions if needed (e.g. role).
Fixes #10169
Fixes #10364
Fixes #10482
Fixes #10630
Fixes #10656
Fixes #10694
Fixes #10918
Fixes #11262
Fixes #11439
Fixes #11453
Fixes #11457
Fixes #11466
Fixes #11540
Fixes #11551
Fixes #11556
Fixes #11654
Fixes #11661
Fixes #11663
Fixes #11664
Fixes #11669
Fixes #11671
Fixes #11807
Fixes #11808
Fixes #11809
Fixes #11815
Fixes #11840
Fixes #11853
Fixes #11860
Fixes #11883
Fixes #11904
Fixes #7388
Fixes #7425
Fixes #7436
Fixes #7544
Fixes #7882
Fixes #7960
Fixes #8255
Fixes #8307
Fixes #8863
Fixes #8927
Fixes #9192
Fixes #9324
2023-12-24 13:11:05 -05:00
|
|
|
cm := qt.Commentf("File: %s Match %s", filename, m)
|
2022-02-09 07:41:04 -05:00
|
|
|
lines := strings.Split(m, "\n")
|
|
|
|
for _, match := range lines {
|
|
|
|
match = strings.TrimSpace(match)
|
|
|
|
if match == "" || strings.HasPrefix(match, "#") {
|
|
|
|
continue
|
|
|
|
}
|
2023-07-01 04:37:38 -04:00
|
|
|
var negate bool
|
|
|
|
if strings.HasPrefix(match, "! ") {
|
|
|
|
negate = true
|
|
|
|
match = strings.TrimPrefix(match, "! ")
|
|
|
|
}
|
|
|
|
if negate {
|
all: Rework page store, add a dynacache, improve partial rebuilds, and some general spring cleaning
There are some breaking changes in this commit, see #11455.
Closes #11455
Closes #11549
This fixes a set of bugs (see issue list) and it is also paying some technical debt accumulated over the years. We now build with Staticcheck enabled in the CI build.
The performance should be about the same as before for regular sized Hugo sites, but it should perform and scale much better to larger data sets, as objects that uses lots of memory (e.g. rendered Markdown, big JSON files read into maps with transform.Unmarshal etc.) will now get automatically garbage collected if needed. Performance on partial rebuilds when running the server in fast render mode should be the same, but the change detection should be much more accurate.
A list of the notable new features:
* A new dependency tracker that covers (almost) all of Hugo's API and is used to do fine grained partial rebuilds when running the server.
* A new and simpler tree document store which allows fast lookups and prefix-walking in all dimensions (e.g. language) concurrently.
* You can now configure an upper memory limit allowing for much larger data sets and/or running on lower specced PCs.
We have lifted the "no resources in sub folders" restriction for branch bundles (e.g. sections).
Memory Limit
* Hugos will, by default, set aside a quarter of the total system memory, but you can set this via the OS environment variable HUGO_MEMORYLIMIT (in gigabytes). This is backed by a partitioned LRU cache used throughout Hugo. A cache that gets dynamically resized in low memory situations, allowing Go's Garbage Collector to free the memory.
New Dependency Tracker: Hugo has had a rule based coarse grained approach to server rebuilds that has worked mostly pretty well, but there have been some surprises (e.g. stale content). This is now revamped with a new dependency tracker that can quickly calculate the delta given a changed resource (e.g. a content file, template, JS file etc.). This handles transitive relations, e.g. $page -> js.Build -> JS import, or $page1.Content -> render hook -> site.GetPage -> $page2.Title, or $page1.Content -> shortcode -> partial -> site.RegularPages -> $page2.Content -> shortcode ..., and should also handle changes to aggregated values (e.g. site.Lastmod) effectively.
This covers all of Hugo's API with 2 known exceptions (a list that may not be fully exhaustive):
Changes to files loaded with template func os.ReadFile may not be handled correctly. We recommend loading resources with resources.Get
Changes to Hugo objects (e.g. Page) passed in the template context to lang.Translate may not be detected correctly. We recommend having simple i18n templates without too much data context passed in other than simple types such as strings and numbers.
Note that the cachebuster configuration (when A changes then rebuild B) works well with the above, but we recommend that you revise that configuration, as it in most situations should not be needed. One example where it is still needed is with TailwindCSS and using changes to hugo_stats.json to trigger new CSS rebuilds.
Document Store: Previously, a little simplified, we split the document store (where we store pages and resources) in a tree per language. This worked pretty well, but the structure made some operations harder than they needed to be. We have now restructured it into one Radix tree for all languages. Internally the language is considered to be a dimension of that tree, and the tree can be viewed in all dimensions concurrently. This makes some operations re. language simpler (e.g. finding translations is just a slice range), but the idea is that it should also be relatively inexpensive to add more dimensions if needed (e.g. role).
Fixes #10169
Fixes #10364
Fixes #10482
Fixes #10630
Fixes #10656
Fixes #10694
Fixes #10918
Fixes #11262
Fixes #11439
Fixes #11453
Fixes #11457
Fixes #11466
Fixes #11540
Fixes #11551
Fixes #11556
Fixes #11654
Fixes #11661
Fixes #11663
Fixes #11664
Fixes #11669
Fixes #11671
Fixes #11807
Fixes #11808
Fixes #11809
Fixes #11815
Fixes #11840
Fixes #11853
Fixes #11860
Fixes #11883
Fixes #11904
Fixes #7388
Fixes #7425
Fixes #7436
Fixes #7544
Fixes #7882
Fixes #7960
Fixes #8255
Fixes #8307
Fixes #8863
Fixes #8927
Fixes #9192
Fixes #9324
2023-12-24 13:11:05 -05:00
|
|
|
s.Assert(content, qt.Not(qt.Contains), match, cm)
|
2023-07-01 04:37:38 -04:00
|
|
|
continue
|
|
|
|
}
|
all: Rework page store, add a dynacache, improve partial rebuilds, and some general spring cleaning
There are some breaking changes in this commit, see #11455.
Closes #11455
Closes #11549
This fixes a set of bugs (see issue list) and it is also paying some technical debt accumulated over the years. We now build with Staticcheck enabled in the CI build.
The performance should be about the same as before for regular sized Hugo sites, but it should perform and scale much better to larger data sets, as objects that uses lots of memory (e.g. rendered Markdown, big JSON files read into maps with transform.Unmarshal etc.) will now get automatically garbage collected if needed. Performance on partial rebuilds when running the server in fast render mode should be the same, but the change detection should be much more accurate.
A list of the notable new features:
* A new dependency tracker that covers (almost) all of Hugo's API and is used to do fine grained partial rebuilds when running the server.
* A new and simpler tree document store which allows fast lookups and prefix-walking in all dimensions (e.g. language) concurrently.
* You can now configure an upper memory limit allowing for much larger data sets and/or running on lower specced PCs.
We have lifted the "no resources in sub folders" restriction for branch bundles (e.g. sections).
Memory Limit
* Hugos will, by default, set aside a quarter of the total system memory, but you can set this via the OS environment variable HUGO_MEMORYLIMIT (in gigabytes). This is backed by a partitioned LRU cache used throughout Hugo. A cache that gets dynamically resized in low memory situations, allowing Go's Garbage Collector to free the memory.
New Dependency Tracker: Hugo has had a rule based coarse grained approach to server rebuilds that has worked mostly pretty well, but there have been some surprises (e.g. stale content). This is now revamped with a new dependency tracker that can quickly calculate the delta given a changed resource (e.g. a content file, template, JS file etc.). This handles transitive relations, e.g. $page -> js.Build -> JS import, or $page1.Content -> render hook -> site.GetPage -> $page2.Title, or $page1.Content -> shortcode -> partial -> site.RegularPages -> $page2.Content -> shortcode ..., and should also handle changes to aggregated values (e.g. site.Lastmod) effectively.
This covers all of Hugo's API with 2 known exceptions (a list that may not be fully exhaustive):
Changes to files loaded with template func os.ReadFile may not be handled correctly. We recommend loading resources with resources.Get
Changes to Hugo objects (e.g. Page) passed in the template context to lang.Translate may not be detected correctly. We recommend having simple i18n templates without too much data context passed in other than simple types such as strings and numbers.
Note that the cachebuster configuration (when A changes then rebuild B) works well with the above, but we recommend that you revise that configuration, as it in most situations should not be needed. One example where it is still needed is with TailwindCSS and using changes to hugo_stats.json to trigger new CSS rebuilds.
Document Store: Previously, a little simplified, we split the document store (where we store pages and resources) in a tree per language. This worked pretty well, but the structure made some operations harder than they needed to be. We have now restructured it into one Radix tree for all languages. Internally the language is considered to be a dimension of that tree, and the tree can be viewed in all dimensions concurrently. This makes some operations re. language simpler (e.g. finding translations is just a slice range), but the idea is that it should also be relatively inexpensive to add more dimensions if needed (e.g. role).
Fixes #10169
Fixes #10364
Fixes #10482
Fixes #10630
Fixes #10656
Fixes #10694
Fixes #10918
Fixes #11262
Fixes #11439
Fixes #11453
Fixes #11457
Fixes #11466
Fixes #11540
Fixes #11551
Fixes #11556
Fixes #11654
Fixes #11661
Fixes #11663
Fixes #11664
Fixes #11669
Fixes #11671
Fixes #11807
Fixes #11808
Fixes #11809
Fixes #11815
Fixes #11840
Fixes #11853
Fixes #11860
Fixes #11883
Fixes #11904
Fixes #7388
Fixes #7425
Fixes #7436
Fixes #7544
Fixes #7882
Fixes #7960
Fixes #8255
Fixes #8307
Fixes #8863
Fixes #8927
Fixes #9192
Fixes #9324
2023-12-24 13:11:05 -05:00
|
|
|
s.Assert(content, qt.Contains, match, cm)
|
2022-02-09 07:41:04 -05:00
|
|
|
}
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
2022-03-11 07:13:18 -05:00
|
|
|
func (s *IntegrationTestBuilder) AssertFileContentExact(filename string, matches ...string) {
|
|
|
|
s.Helper()
|
|
|
|
content := s.FileContent(filename)
|
|
|
|
for _, m := range matches {
|
|
|
|
s.Assert(content, qt.Contains, m, qt.Commentf(m))
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
2024-02-07 04:30:32 -05:00
|
|
|
func (s *IntegrationTestBuilder) AssertPublishDir(matches ...string) {
|
|
|
|
s.Helper()
|
|
|
|
var buff bytes.Buffer
|
|
|
|
helpers.PrintFs(s.H.Fs.PublishDir, "", &buff)
|
|
|
|
printFsLines := strings.Split(buff.String(), "\n")
|
|
|
|
sort.Strings(printFsLines)
|
|
|
|
content := strings.TrimSpace((strings.Join(printFsLines, "\n")))
|
|
|
|
for _, m := range matches {
|
|
|
|
cm := qt.Commentf("Match: %q\nIn:\n%s", m, content)
|
|
|
|
lines := strings.Split(m, "\n")
|
|
|
|
for _, match := range lines {
|
|
|
|
match = strings.TrimSpace(match)
|
|
|
|
var negate bool
|
|
|
|
if strings.HasPrefix(match, "! ") {
|
|
|
|
negate = true
|
|
|
|
match = strings.TrimPrefix(match, "! ")
|
|
|
|
}
|
|
|
|
if negate {
|
|
|
|
s.Assert(content, qt.Not(qt.Contains), match, cm)
|
|
|
|
continue
|
|
|
|
}
|
|
|
|
s.Assert(content, qt.Contains, match, cm)
|
|
|
|
}
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
2023-11-02 04:50:18 -04:00
|
|
|
func (s *IntegrationTestBuilder) AssertFileExists(filename string, b bool) {
|
|
|
|
checker := qt.IsNil
|
|
|
|
if !b {
|
|
|
|
checker = qt.IsNotNil
|
|
|
|
}
|
|
|
|
_, err := s.fs.WorkingDirReadOnly.Stat(filename)
|
|
|
|
if !herrors.IsNotExist(err) {
|
|
|
|
s.Assert(err, qt.IsNil)
|
|
|
|
}
|
|
|
|
s.Assert(err, checker)
|
|
|
|
}
|
|
|
|
|
2022-05-15 05:40:34 -04:00
|
|
|
func (s *IntegrationTestBuilder) AssertIsFileError(err error) herrors.FileError {
|
2022-05-02 10:07:52 -04:00
|
|
|
s.Assert(err, qt.ErrorAs, new(herrors.FileError))
|
2022-05-15 05:40:34 -04:00
|
|
|
return herrors.UnwrapFileError(err)
|
2022-02-09 07:41:04 -05:00
|
|
|
}
|
|
|
|
|
|
|
|
func (s *IntegrationTestBuilder) AssertRenderCountContent(count int) {
|
|
|
|
s.Helper()
|
all: Rework page store, add a dynacache, improve partial rebuilds, and some general spring cleaning
There are some breaking changes in this commit, see #11455.
Closes #11455
Closes #11549
This fixes a set of bugs (see issue list) and it is also paying some technical debt accumulated over the years. We now build with Staticcheck enabled in the CI build.
The performance should be about the same as before for regular sized Hugo sites, but it should perform and scale much better to larger data sets, as objects that uses lots of memory (e.g. rendered Markdown, big JSON files read into maps with transform.Unmarshal etc.) will now get automatically garbage collected if needed. Performance on partial rebuilds when running the server in fast render mode should be the same, but the change detection should be much more accurate.
A list of the notable new features:
* A new dependency tracker that covers (almost) all of Hugo's API and is used to do fine grained partial rebuilds when running the server.
* A new and simpler tree document store which allows fast lookups and prefix-walking in all dimensions (e.g. language) concurrently.
* You can now configure an upper memory limit allowing for much larger data sets and/or running on lower specced PCs.
We have lifted the "no resources in sub folders" restriction for branch bundles (e.g. sections).
Memory Limit
* Hugos will, by default, set aside a quarter of the total system memory, but you can set this via the OS environment variable HUGO_MEMORYLIMIT (in gigabytes). This is backed by a partitioned LRU cache used throughout Hugo. A cache that gets dynamically resized in low memory situations, allowing Go's Garbage Collector to free the memory.
New Dependency Tracker: Hugo has had a rule based coarse grained approach to server rebuilds that has worked mostly pretty well, but there have been some surprises (e.g. stale content). This is now revamped with a new dependency tracker that can quickly calculate the delta given a changed resource (e.g. a content file, template, JS file etc.). This handles transitive relations, e.g. $page -> js.Build -> JS import, or $page1.Content -> render hook -> site.GetPage -> $page2.Title, or $page1.Content -> shortcode -> partial -> site.RegularPages -> $page2.Content -> shortcode ..., and should also handle changes to aggregated values (e.g. site.Lastmod) effectively.
This covers all of Hugo's API with 2 known exceptions (a list that may not be fully exhaustive):
Changes to files loaded with template func os.ReadFile may not be handled correctly. We recommend loading resources with resources.Get
Changes to Hugo objects (e.g. Page) passed in the template context to lang.Translate may not be detected correctly. We recommend having simple i18n templates without too much data context passed in other than simple types such as strings and numbers.
Note that the cachebuster configuration (when A changes then rebuild B) works well with the above, but we recommend that you revise that configuration, as it in most situations should not be needed. One example where it is still needed is with TailwindCSS and using changes to hugo_stats.json to trigger new CSS rebuilds.
Document Store: Previously, a little simplified, we split the document store (where we store pages and resources) in a tree per language. This worked pretty well, but the structure made some operations harder than they needed to be. We have now restructured it into one Radix tree for all languages. Internally the language is considered to be a dimension of that tree, and the tree can be viewed in all dimensions concurrently. This makes some operations re. language simpler (e.g. finding translations is just a slice range), but the idea is that it should also be relatively inexpensive to add more dimensions if needed (e.g. role).
Fixes #10169
Fixes #10364
Fixes #10482
Fixes #10630
Fixes #10656
Fixes #10694
Fixes #10918
Fixes #11262
Fixes #11439
Fixes #11453
Fixes #11457
Fixes #11466
Fixes #11540
Fixes #11551
Fixes #11556
Fixes #11654
Fixes #11661
Fixes #11663
Fixes #11664
Fixes #11669
Fixes #11671
Fixes #11807
Fixes #11808
Fixes #11809
Fixes #11815
Fixes #11840
Fixes #11853
Fixes #11860
Fixes #11883
Fixes #11904
Fixes #7388
Fixes #7425
Fixes #7436
Fixes #7544
Fixes #7882
Fixes #7960
Fixes #8255
Fixes #8307
Fixes #8863
Fixes #8927
Fixes #9192
Fixes #9324
2023-12-24 13:11:05 -05:00
|
|
|
s.Assert(s.counters.contentRenderCounter.Load(), qt.Equals, uint64(count))
|
2022-02-09 07:41:04 -05:00
|
|
|
}
|
|
|
|
|
|
|
|
func (s *IntegrationTestBuilder) AssertRenderCountPage(count int) {
|
|
|
|
s.Helper()
|
all: Rework page store, add a dynacache, improve partial rebuilds, and some general spring cleaning
There are some breaking changes in this commit, see #11455.
Closes #11455
Closes #11549
This fixes a set of bugs (see issue list) and it is also paying some technical debt accumulated over the years. We now build with Staticcheck enabled in the CI build.
The performance should be about the same as before for regular sized Hugo sites, but it should perform and scale much better to larger data sets, as objects that uses lots of memory (e.g. rendered Markdown, big JSON files read into maps with transform.Unmarshal etc.) will now get automatically garbage collected if needed. Performance on partial rebuilds when running the server in fast render mode should be the same, but the change detection should be much more accurate.
A list of the notable new features:
* A new dependency tracker that covers (almost) all of Hugo's API and is used to do fine grained partial rebuilds when running the server.
* A new and simpler tree document store which allows fast lookups and prefix-walking in all dimensions (e.g. language) concurrently.
* You can now configure an upper memory limit allowing for much larger data sets and/or running on lower specced PCs.
We have lifted the "no resources in sub folders" restriction for branch bundles (e.g. sections).
Memory Limit
* Hugos will, by default, set aside a quarter of the total system memory, but you can set this via the OS environment variable HUGO_MEMORYLIMIT (in gigabytes). This is backed by a partitioned LRU cache used throughout Hugo. A cache that gets dynamically resized in low memory situations, allowing Go's Garbage Collector to free the memory.
New Dependency Tracker: Hugo has had a rule based coarse grained approach to server rebuilds that has worked mostly pretty well, but there have been some surprises (e.g. stale content). This is now revamped with a new dependency tracker that can quickly calculate the delta given a changed resource (e.g. a content file, template, JS file etc.). This handles transitive relations, e.g. $page -> js.Build -> JS import, or $page1.Content -> render hook -> site.GetPage -> $page2.Title, or $page1.Content -> shortcode -> partial -> site.RegularPages -> $page2.Content -> shortcode ..., and should also handle changes to aggregated values (e.g. site.Lastmod) effectively.
This covers all of Hugo's API with 2 known exceptions (a list that may not be fully exhaustive):
Changes to files loaded with template func os.ReadFile may not be handled correctly. We recommend loading resources with resources.Get
Changes to Hugo objects (e.g. Page) passed in the template context to lang.Translate may not be detected correctly. We recommend having simple i18n templates without too much data context passed in other than simple types such as strings and numbers.
Note that the cachebuster configuration (when A changes then rebuild B) works well with the above, but we recommend that you revise that configuration, as it in most situations should not be needed. One example where it is still needed is with TailwindCSS and using changes to hugo_stats.json to trigger new CSS rebuilds.
Document Store: Previously, a little simplified, we split the document store (where we store pages and resources) in a tree per language. This worked pretty well, but the structure made some operations harder than they needed to be. We have now restructured it into one Radix tree for all languages. Internally the language is considered to be a dimension of that tree, and the tree can be viewed in all dimensions concurrently. This makes some operations re. language simpler (e.g. finding translations is just a slice range), but the idea is that it should also be relatively inexpensive to add more dimensions if needed (e.g. role).
Fixes #10169
Fixes #10364
Fixes #10482
Fixes #10630
Fixes #10656
Fixes #10694
Fixes #10918
Fixes #11262
Fixes #11439
Fixes #11453
Fixes #11457
Fixes #11466
Fixes #11540
Fixes #11551
Fixes #11556
Fixes #11654
Fixes #11661
Fixes #11663
Fixes #11664
Fixes #11669
Fixes #11671
Fixes #11807
Fixes #11808
Fixes #11809
Fixes #11815
Fixes #11840
Fixes #11853
Fixes #11860
Fixes #11883
Fixes #11904
Fixes #7388
Fixes #7425
Fixes #7436
Fixes #7544
Fixes #7882
Fixes #7960
Fixes #8255
Fixes #8307
Fixes #8863
Fixes #8927
Fixes #9192
Fixes #9324
2023-12-24 13:11:05 -05:00
|
|
|
s.Assert(s.counters.pageRenderCounter.Load(), qt.Equals, uint64(count))
|
|
|
|
}
|
|
|
|
|
|
|
|
func (s *IntegrationTestBuilder) AssertRenderCountPageBetween(from, to int) {
|
|
|
|
s.Helper()
|
|
|
|
i := int(s.counters.pageRenderCounter.Load())
|
|
|
|
s.Assert(i >= from && i <= to, qt.IsTrue)
|
2022-02-09 07:41:04 -05:00
|
|
|
}
|
|
|
|
|
|
|
|
func (s *IntegrationTestBuilder) Build() *IntegrationTestBuilder {
|
|
|
|
s.Helper()
|
|
|
|
_, err := s.BuildE()
|
2022-02-17 07:04:00 -05:00
|
|
|
if s.Cfg.Verbose || err != nil {
|
2022-02-09 07:41:04 -05:00
|
|
|
fmt.Println(s.logBuff.String())
|
all: Rework page store, add a dynacache, improve partial rebuilds, and some general spring cleaning
There are some breaking changes in this commit, see #11455.
Closes #11455
Closes #11549
This fixes a set of bugs (see issue list) and it is also paying some technical debt accumulated over the years. We now build with Staticcheck enabled in the CI build.
The performance should be about the same as before for regular sized Hugo sites, but it should perform and scale much better to larger data sets, as objects that uses lots of memory (e.g. rendered Markdown, big JSON files read into maps with transform.Unmarshal etc.) will now get automatically garbage collected if needed. Performance on partial rebuilds when running the server in fast render mode should be the same, but the change detection should be much more accurate.
A list of the notable new features:
* A new dependency tracker that covers (almost) all of Hugo's API and is used to do fine grained partial rebuilds when running the server.
* A new and simpler tree document store which allows fast lookups and prefix-walking in all dimensions (e.g. language) concurrently.
* You can now configure an upper memory limit allowing for much larger data sets and/or running on lower specced PCs.
We have lifted the "no resources in sub folders" restriction for branch bundles (e.g. sections).
Memory Limit
* Hugos will, by default, set aside a quarter of the total system memory, but you can set this via the OS environment variable HUGO_MEMORYLIMIT (in gigabytes). This is backed by a partitioned LRU cache used throughout Hugo. A cache that gets dynamically resized in low memory situations, allowing Go's Garbage Collector to free the memory.
New Dependency Tracker: Hugo has had a rule based coarse grained approach to server rebuilds that has worked mostly pretty well, but there have been some surprises (e.g. stale content). This is now revamped with a new dependency tracker that can quickly calculate the delta given a changed resource (e.g. a content file, template, JS file etc.). This handles transitive relations, e.g. $page -> js.Build -> JS import, or $page1.Content -> render hook -> site.GetPage -> $page2.Title, or $page1.Content -> shortcode -> partial -> site.RegularPages -> $page2.Content -> shortcode ..., and should also handle changes to aggregated values (e.g. site.Lastmod) effectively.
This covers all of Hugo's API with 2 known exceptions (a list that may not be fully exhaustive):
Changes to files loaded with template func os.ReadFile may not be handled correctly. We recommend loading resources with resources.Get
Changes to Hugo objects (e.g. Page) passed in the template context to lang.Translate may not be detected correctly. We recommend having simple i18n templates without too much data context passed in other than simple types such as strings and numbers.
Note that the cachebuster configuration (when A changes then rebuild B) works well with the above, but we recommend that you revise that configuration, as it in most situations should not be needed. One example where it is still needed is with TailwindCSS and using changes to hugo_stats.json to trigger new CSS rebuilds.
Document Store: Previously, a little simplified, we split the document store (where we store pages and resources) in a tree per language. This worked pretty well, but the structure made some operations harder than they needed to be. We have now restructured it into one Radix tree for all languages. Internally the language is considered to be a dimension of that tree, and the tree can be viewed in all dimensions concurrently. This makes some operations re. language simpler (e.g. finding translations is just a slice range), but the idea is that it should also be relatively inexpensive to add more dimensions if needed (e.g. role).
Fixes #10169
Fixes #10364
Fixes #10482
Fixes #10630
Fixes #10656
Fixes #10694
Fixes #10918
Fixes #11262
Fixes #11439
Fixes #11453
Fixes #11457
Fixes #11466
Fixes #11540
Fixes #11551
Fixes #11556
Fixes #11654
Fixes #11661
Fixes #11663
Fixes #11664
Fixes #11669
Fixes #11671
Fixes #11807
Fixes #11808
Fixes #11809
Fixes #11815
Fixes #11840
Fixes #11853
Fixes #11860
Fixes #11883
Fixes #11904
Fixes #7388
Fixes #7425
Fixes #7436
Fixes #7544
Fixes #7882
Fixes #7960
Fixes #8255
Fixes #8307
Fixes #8863
Fixes #8927
Fixes #9192
Fixes #9324
2023-12-24 13:11:05 -05:00
|
|
|
if s.H != nil && err == nil {
|
|
|
|
for _, s := range s.H.Sites {
|
|
|
|
m := s.pageMap
|
|
|
|
var buff bytes.Buffer
|
|
|
|
fmt.Fprintf(&buff, "PageMap for site %q\n\n", s.Language().Lang)
|
|
|
|
m.debugPrint("", 999, &buff)
|
|
|
|
fmt.Println(buff.String())
|
|
|
|
}
|
|
|
|
}
|
|
|
|
} else if s.Cfg.LogLevel <= logg.LevelDebug {
|
|
|
|
fmt.Println(s.logBuff.String())
|
2022-02-09 07:41:04 -05:00
|
|
|
}
|
2023-01-04 12:24:36 -05:00
|
|
|
s.Assert(err, qt.IsNil)
|
2023-03-04 05:09:03 -05:00
|
|
|
if s.Cfg.RunGC {
|
|
|
|
s.GCCount, err = s.H.GC()
|
all: Rework page store, add a dynacache, improve partial rebuilds, and some general spring cleaning
There are some breaking changes in this commit, see #11455.
Closes #11455
Closes #11549
This fixes a set of bugs (see issue list) and it is also paying some technical debt accumulated over the years. We now build with Staticcheck enabled in the CI build.
The performance should be about the same as before for regular sized Hugo sites, but it should perform and scale much better to larger data sets, as objects that uses lots of memory (e.g. rendered Markdown, big JSON files read into maps with transform.Unmarshal etc.) will now get automatically garbage collected if needed. Performance on partial rebuilds when running the server in fast render mode should be the same, but the change detection should be much more accurate.
A list of the notable new features:
* A new dependency tracker that covers (almost) all of Hugo's API and is used to do fine grained partial rebuilds when running the server.
* A new and simpler tree document store which allows fast lookups and prefix-walking in all dimensions (e.g. language) concurrently.
* You can now configure an upper memory limit allowing for much larger data sets and/or running on lower specced PCs.
We have lifted the "no resources in sub folders" restriction for branch bundles (e.g. sections).
Memory Limit
* Hugos will, by default, set aside a quarter of the total system memory, but you can set this via the OS environment variable HUGO_MEMORYLIMIT (in gigabytes). This is backed by a partitioned LRU cache used throughout Hugo. A cache that gets dynamically resized in low memory situations, allowing Go's Garbage Collector to free the memory.
New Dependency Tracker: Hugo has had a rule based coarse grained approach to server rebuilds that has worked mostly pretty well, but there have been some surprises (e.g. stale content). This is now revamped with a new dependency tracker that can quickly calculate the delta given a changed resource (e.g. a content file, template, JS file etc.). This handles transitive relations, e.g. $page -> js.Build -> JS import, or $page1.Content -> render hook -> site.GetPage -> $page2.Title, or $page1.Content -> shortcode -> partial -> site.RegularPages -> $page2.Content -> shortcode ..., and should also handle changes to aggregated values (e.g. site.Lastmod) effectively.
This covers all of Hugo's API with 2 known exceptions (a list that may not be fully exhaustive):
Changes to files loaded with template func os.ReadFile may not be handled correctly. We recommend loading resources with resources.Get
Changes to Hugo objects (e.g. Page) passed in the template context to lang.Translate may not be detected correctly. We recommend having simple i18n templates without too much data context passed in other than simple types such as strings and numbers.
Note that the cachebuster configuration (when A changes then rebuild B) works well with the above, but we recommend that you revise that configuration, as it in most situations should not be needed. One example where it is still needed is with TailwindCSS and using changes to hugo_stats.json to trigger new CSS rebuilds.
Document Store: Previously, a little simplified, we split the document store (where we store pages and resources) in a tree per language. This worked pretty well, but the structure made some operations harder than they needed to be. We have now restructured it into one Radix tree for all languages. Internally the language is considered to be a dimension of that tree, and the tree can be viewed in all dimensions concurrently. This makes some operations re. language simpler (e.g. finding translations is just a slice range), but the idea is that it should also be relatively inexpensive to add more dimensions if needed (e.g. role).
Fixes #10169
Fixes #10364
Fixes #10482
Fixes #10630
Fixes #10656
Fixes #10694
Fixes #10918
Fixes #11262
Fixes #11439
Fixes #11453
Fixes #11457
Fixes #11466
Fixes #11540
Fixes #11551
Fixes #11556
Fixes #11654
Fixes #11661
Fixes #11663
Fixes #11664
Fixes #11669
Fixes #11671
Fixes #11807
Fixes #11808
Fixes #11809
Fixes #11815
Fixes #11840
Fixes #11853
Fixes #11860
Fixes #11883
Fixes #11904
Fixes #7388
Fixes #7425
Fixes #7436
Fixes #7544
Fixes #7882
Fixes #7960
Fixes #8255
Fixes #8307
Fixes #8863
Fixes #8927
Fixes #9192
Fixes #9324
2023-12-24 13:11:05 -05:00
|
|
|
s.Assert(err, qt.IsNil)
|
2023-03-04 05:09:03 -05:00
|
|
|
}
|
2023-01-04 12:24:36 -05:00
|
|
|
|
2022-02-09 07:41:04 -05:00
|
|
|
return s
|
|
|
|
}
|
|
|
|
|
2023-06-28 04:27:39 -04:00
|
|
|
func (s *IntegrationTestBuilder) LogString() string {
|
|
|
|
return s.logBuff.String()
|
|
|
|
}
|
|
|
|
|
2022-02-09 07:41:04 -05:00
|
|
|
func (s *IntegrationTestBuilder) BuildE() (*IntegrationTestBuilder, error) {
|
|
|
|
s.Helper()
|
2022-06-04 11:39:34 -04:00
|
|
|
if err := s.initBuilder(); err != nil {
|
|
|
|
return s, err
|
|
|
|
}
|
|
|
|
|
2023-05-15 02:39:52 -04:00
|
|
|
err := s.build(s.Cfg.BuildCfg)
|
2022-02-09 07:41:04 -05:00
|
|
|
return s, err
|
|
|
|
}
|
|
|
|
|
2023-07-28 06:04:03 -04:00
|
|
|
func (s *IntegrationTestBuilder) Init() *IntegrationTestBuilder {
|
|
|
|
if err := s.initBuilder(); err != nil {
|
|
|
|
s.Fatalf("Failed to init builder: %s", err)
|
|
|
|
}
|
|
|
|
return s
|
|
|
|
}
|
|
|
|
|
2022-02-09 07:41:04 -05:00
|
|
|
type IntegrationTestDebugConfig struct {
|
|
|
|
Out io.Writer
|
|
|
|
|
|
|
|
PrintDestinationFs bool
|
|
|
|
PrintPagemap bool
|
|
|
|
|
|
|
|
PrefixDestinationFs string
|
|
|
|
PrefixPagemap string
|
|
|
|
}
|
|
|
|
|
all: Rework page store, add a dynacache, improve partial rebuilds, and some general spring cleaning
There are some breaking changes in this commit, see #11455.
Closes #11455
Closes #11549
This fixes a set of bugs (see issue list) and it is also paying some technical debt accumulated over the years. We now build with Staticcheck enabled in the CI build.
The performance should be about the same as before for regular sized Hugo sites, but it should perform and scale much better to larger data sets, as objects that uses lots of memory (e.g. rendered Markdown, big JSON files read into maps with transform.Unmarshal etc.) will now get automatically garbage collected if needed. Performance on partial rebuilds when running the server in fast render mode should be the same, but the change detection should be much more accurate.
A list of the notable new features:
* A new dependency tracker that covers (almost) all of Hugo's API and is used to do fine grained partial rebuilds when running the server.
* A new and simpler tree document store which allows fast lookups and prefix-walking in all dimensions (e.g. language) concurrently.
* You can now configure an upper memory limit allowing for much larger data sets and/or running on lower specced PCs.
We have lifted the "no resources in sub folders" restriction for branch bundles (e.g. sections).
Memory Limit
* Hugos will, by default, set aside a quarter of the total system memory, but you can set this via the OS environment variable HUGO_MEMORYLIMIT (in gigabytes). This is backed by a partitioned LRU cache used throughout Hugo. A cache that gets dynamically resized in low memory situations, allowing Go's Garbage Collector to free the memory.
New Dependency Tracker: Hugo has had a rule based coarse grained approach to server rebuilds that has worked mostly pretty well, but there have been some surprises (e.g. stale content). This is now revamped with a new dependency tracker that can quickly calculate the delta given a changed resource (e.g. a content file, template, JS file etc.). This handles transitive relations, e.g. $page -> js.Build -> JS import, or $page1.Content -> render hook -> site.GetPage -> $page2.Title, or $page1.Content -> shortcode -> partial -> site.RegularPages -> $page2.Content -> shortcode ..., and should also handle changes to aggregated values (e.g. site.Lastmod) effectively.
This covers all of Hugo's API with 2 known exceptions (a list that may not be fully exhaustive):
Changes to files loaded with template func os.ReadFile may not be handled correctly. We recommend loading resources with resources.Get
Changes to Hugo objects (e.g. Page) passed in the template context to lang.Translate may not be detected correctly. We recommend having simple i18n templates without too much data context passed in other than simple types such as strings and numbers.
Note that the cachebuster configuration (when A changes then rebuild B) works well with the above, but we recommend that you revise that configuration, as it in most situations should not be needed. One example where it is still needed is with TailwindCSS and using changes to hugo_stats.json to trigger new CSS rebuilds.
Document Store: Previously, a little simplified, we split the document store (where we store pages and resources) in a tree per language. This worked pretty well, but the structure made some operations harder than they needed to be. We have now restructured it into one Radix tree for all languages. Internally the language is considered to be a dimension of that tree, and the tree can be viewed in all dimensions concurrently. This makes some operations re. language simpler (e.g. finding translations is just a slice range), but the idea is that it should also be relatively inexpensive to add more dimensions if needed (e.g. role).
Fixes #10169
Fixes #10364
Fixes #10482
Fixes #10630
Fixes #10656
Fixes #10694
Fixes #10918
Fixes #11262
Fixes #11439
Fixes #11453
Fixes #11457
Fixes #11466
Fixes #11540
Fixes #11551
Fixes #11556
Fixes #11654
Fixes #11661
Fixes #11663
Fixes #11664
Fixes #11669
Fixes #11671
Fixes #11807
Fixes #11808
Fixes #11809
Fixes #11815
Fixes #11840
Fixes #11853
Fixes #11860
Fixes #11883
Fixes #11904
Fixes #7388
Fixes #7425
Fixes #7436
Fixes #7544
Fixes #7882
Fixes #7960
Fixes #8255
Fixes #8307
Fixes #8863
Fixes #8927
Fixes #9192
Fixes #9324
2023-12-24 13:11:05 -05:00
|
|
|
func (s *IntegrationTestBuilder) EditFileReplaceAll(filename, old, new string) *IntegrationTestBuilder {
|
|
|
|
return s.EditFileReplaceFunc(filename, func(s string) string {
|
|
|
|
return strings.ReplaceAll(s, old, new)
|
|
|
|
})
|
|
|
|
}
|
|
|
|
|
|
|
|
func (s *IntegrationTestBuilder) EditFileReplaceFunc(filename string, replacementFunc func(s string) string) *IntegrationTestBuilder {
|
2022-02-09 07:41:04 -05:00
|
|
|
absFilename := s.absFilename(filename)
|
|
|
|
b, err := afero.ReadFile(s.fs.Source, absFilename)
|
|
|
|
s.Assert(err, qt.IsNil)
|
|
|
|
s.changedFiles = append(s.changedFiles, absFilename)
|
|
|
|
oldContent := string(b)
|
|
|
|
s.writeSource(absFilename, replacementFunc(oldContent))
|
|
|
|
return s
|
|
|
|
}
|
|
|
|
|
|
|
|
func (s *IntegrationTestBuilder) EditFiles(filenameContent ...string) *IntegrationTestBuilder {
|
|
|
|
for i := 0; i < len(filenameContent); i += 2 {
|
|
|
|
filename, content := filepath.FromSlash(filenameContent[i]), filenameContent[i+1]
|
|
|
|
absFilename := s.absFilename(filename)
|
|
|
|
s.changedFiles = append(s.changedFiles, absFilename)
|
|
|
|
s.writeSource(absFilename, content)
|
|
|
|
}
|
|
|
|
return s
|
|
|
|
}
|
|
|
|
|
|
|
|
func (s *IntegrationTestBuilder) AddFiles(filenameContent ...string) *IntegrationTestBuilder {
|
|
|
|
for i := 0; i < len(filenameContent); i += 2 {
|
|
|
|
filename, content := filepath.FromSlash(filenameContent[i]), filenameContent[i+1]
|
|
|
|
absFilename := s.absFilename(filename)
|
|
|
|
s.createdFiles = append(s.createdFiles, absFilename)
|
|
|
|
s.writeSource(absFilename, content)
|
|
|
|
}
|
|
|
|
return s
|
|
|
|
}
|
|
|
|
|
|
|
|
func (s *IntegrationTestBuilder) RemoveFiles(filenames ...string) *IntegrationTestBuilder {
|
|
|
|
for _, filename := range filenames {
|
|
|
|
absFilename := s.absFilename(filename)
|
|
|
|
s.removedFiles = append(s.removedFiles, absFilename)
|
|
|
|
s.Assert(s.fs.Source.Remove(absFilename), qt.IsNil)
|
|
|
|
|
|
|
|
}
|
|
|
|
|
|
|
|
return s
|
|
|
|
}
|
|
|
|
|
|
|
|
func (s *IntegrationTestBuilder) RenameFile(old, new string) *IntegrationTestBuilder {
|
|
|
|
absOldFilename := s.absFilename(old)
|
|
|
|
absNewFilename := s.absFilename(new)
|
|
|
|
s.renamedFiles = append(s.renamedFiles, absOldFilename)
|
|
|
|
s.createdFiles = append(s.createdFiles, absNewFilename)
|
2023-10-18 12:54:15 -04:00
|
|
|
s.Assert(s.fs.Source.MkdirAll(filepath.Dir(absNewFilename), 0o777), qt.IsNil)
|
2022-02-09 07:41:04 -05:00
|
|
|
s.Assert(s.fs.Source.Rename(absOldFilename, absNewFilename), qt.IsNil)
|
|
|
|
return s
|
|
|
|
}
|
|
|
|
|
all: Rework page store, add a dynacache, improve partial rebuilds, and some general spring cleaning
There are some breaking changes in this commit, see #11455.
Closes #11455
Closes #11549
This fixes a set of bugs (see issue list) and it is also paying some technical debt accumulated over the years. We now build with Staticcheck enabled in the CI build.
The performance should be about the same as before for regular sized Hugo sites, but it should perform and scale much better to larger data sets, as objects that uses lots of memory (e.g. rendered Markdown, big JSON files read into maps with transform.Unmarshal etc.) will now get automatically garbage collected if needed. Performance on partial rebuilds when running the server in fast render mode should be the same, but the change detection should be much more accurate.
A list of the notable new features:
* A new dependency tracker that covers (almost) all of Hugo's API and is used to do fine grained partial rebuilds when running the server.
* A new and simpler tree document store which allows fast lookups and prefix-walking in all dimensions (e.g. language) concurrently.
* You can now configure an upper memory limit allowing for much larger data sets and/or running on lower specced PCs.
We have lifted the "no resources in sub folders" restriction for branch bundles (e.g. sections).
Memory Limit
* Hugos will, by default, set aside a quarter of the total system memory, but you can set this via the OS environment variable HUGO_MEMORYLIMIT (in gigabytes). This is backed by a partitioned LRU cache used throughout Hugo. A cache that gets dynamically resized in low memory situations, allowing Go's Garbage Collector to free the memory.
New Dependency Tracker: Hugo has had a rule based coarse grained approach to server rebuilds that has worked mostly pretty well, but there have been some surprises (e.g. stale content). This is now revamped with a new dependency tracker that can quickly calculate the delta given a changed resource (e.g. a content file, template, JS file etc.). This handles transitive relations, e.g. $page -> js.Build -> JS import, or $page1.Content -> render hook -> site.GetPage -> $page2.Title, or $page1.Content -> shortcode -> partial -> site.RegularPages -> $page2.Content -> shortcode ..., and should also handle changes to aggregated values (e.g. site.Lastmod) effectively.
This covers all of Hugo's API with 2 known exceptions (a list that may not be fully exhaustive):
Changes to files loaded with template func os.ReadFile may not be handled correctly. We recommend loading resources with resources.Get
Changes to Hugo objects (e.g. Page) passed in the template context to lang.Translate may not be detected correctly. We recommend having simple i18n templates without too much data context passed in other than simple types such as strings and numbers.
Note that the cachebuster configuration (when A changes then rebuild B) works well with the above, but we recommend that you revise that configuration, as it in most situations should not be needed. One example where it is still needed is with TailwindCSS and using changes to hugo_stats.json to trigger new CSS rebuilds.
Document Store: Previously, a little simplified, we split the document store (where we store pages and resources) in a tree per language. This worked pretty well, but the structure made some operations harder than they needed to be. We have now restructured it into one Radix tree for all languages. Internally the language is considered to be a dimension of that tree, and the tree can be viewed in all dimensions concurrently. This makes some operations re. language simpler (e.g. finding translations is just a slice range), but the idea is that it should also be relatively inexpensive to add more dimensions if needed (e.g. role).
Fixes #10169
Fixes #10364
Fixes #10482
Fixes #10630
Fixes #10656
Fixes #10694
Fixes #10918
Fixes #11262
Fixes #11439
Fixes #11453
Fixes #11457
Fixes #11466
Fixes #11540
Fixes #11551
Fixes #11556
Fixes #11654
Fixes #11661
Fixes #11663
Fixes #11664
Fixes #11669
Fixes #11671
Fixes #11807
Fixes #11808
Fixes #11809
Fixes #11815
Fixes #11840
Fixes #11853
Fixes #11860
Fixes #11883
Fixes #11904
Fixes #7388
Fixes #7425
Fixes #7436
Fixes #7544
Fixes #7882
Fixes #7960
Fixes #8255
Fixes #8307
Fixes #8863
Fixes #8927
Fixes #9192
Fixes #9324
2023-12-24 13:11:05 -05:00
|
|
|
func (s *IntegrationTestBuilder) RenameDir(old, new string) *IntegrationTestBuilder {
|
|
|
|
absOldFilename := s.absFilename(old)
|
|
|
|
absNewFilename := s.absFilename(new)
|
|
|
|
s.renamedDirs = append(s.renamedDirs, absOldFilename)
|
|
|
|
s.changedFiles = append(s.changedFiles, absNewFilename)
|
|
|
|
afero.Walk(s.fs.Source, absOldFilename, func(path string, info os.FileInfo, err error) error {
|
|
|
|
if err != nil {
|
|
|
|
return err
|
|
|
|
}
|
|
|
|
if info.IsDir() {
|
|
|
|
return nil
|
|
|
|
}
|
|
|
|
s.createdFiles = append(s.createdFiles, strings.Replace(path, absOldFilename, absNewFilename, 1))
|
|
|
|
return nil
|
|
|
|
})
|
|
|
|
s.Assert(s.fs.Source.MkdirAll(filepath.Dir(absNewFilename), 0o777), qt.IsNil)
|
|
|
|
s.Assert(s.fs.Source.Rename(absOldFilename, absNewFilename), qt.IsNil)
|
|
|
|
return s
|
|
|
|
}
|
|
|
|
|
2022-02-09 07:41:04 -05:00
|
|
|
func (s *IntegrationTestBuilder) FileContent(filename string) string {
|
|
|
|
s.Helper()
|
2022-03-21 04:35:15 -04:00
|
|
|
return s.readWorkingDir(s, s.fs, filepath.FromSlash(filename))
|
2022-02-09 07:41:04 -05:00
|
|
|
}
|
|
|
|
|
2022-06-04 11:39:34 -04:00
|
|
|
func (s *IntegrationTestBuilder) initBuilder() error {
|
|
|
|
var initErr error
|
2022-02-09 07:41:04 -05:00
|
|
|
s.builderInit.Do(func() {
|
|
|
|
var afs afero.Fs
|
|
|
|
if s.Cfg.NeedsOsFS {
|
|
|
|
afs = afero.NewOsFs()
|
|
|
|
} else {
|
|
|
|
afs = afero.NewMemMapFs()
|
|
|
|
}
|
|
|
|
|
|
|
|
if s.Cfg.LogLevel == 0 {
|
all: Rework page store, add a dynacache, improve partial rebuilds, and some general spring cleaning
There are some breaking changes in this commit, see #11455.
Closes #11455
Closes #11549
This fixes a set of bugs (see issue list) and it is also paying some technical debt accumulated over the years. We now build with Staticcheck enabled in the CI build.
The performance should be about the same as before for regular sized Hugo sites, but it should perform and scale much better to larger data sets, as objects that uses lots of memory (e.g. rendered Markdown, big JSON files read into maps with transform.Unmarshal etc.) will now get automatically garbage collected if needed. Performance on partial rebuilds when running the server in fast render mode should be the same, but the change detection should be much more accurate.
A list of the notable new features:
* A new dependency tracker that covers (almost) all of Hugo's API and is used to do fine grained partial rebuilds when running the server.
* A new and simpler tree document store which allows fast lookups and prefix-walking in all dimensions (e.g. language) concurrently.
* You can now configure an upper memory limit allowing for much larger data sets and/or running on lower specced PCs.
We have lifted the "no resources in sub folders" restriction for branch bundles (e.g. sections).
Memory Limit
* Hugos will, by default, set aside a quarter of the total system memory, but you can set this via the OS environment variable HUGO_MEMORYLIMIT (in gigabytes). This is backed by a partitioned LRU cache used throughout Hugo. A cache that gets dynamically resized in low memory situations, allowing Go's Garbage Collector to free the memory.
New Dependency Tracker: Hugo has had a rule based coarse grained approach to server rebuilds that has worked mostly pretty well, but there have been some surprises (e.g. stale content). This is now revamped with a new dependency tracker that can quickly calculate the delta given a changed resource (e.g. a content file, template, JS file etc.). This handles transitive relations, e.g. $page -> js.Build -> JS import, or $page1.Content -> render hook -> site.GetPage -> $page2.Title, or $page1.Content -> shortcode -> partial -> site.RegularPages -> $page2.Content -> shortcode ..., and should also handle changes to aggregated values (e.g. site.Lastmod) effectively.
This covers all of Hugo's API with 2 known exceptions (a list that may not be fully exhaustive):
Changes to files loaded with template func os.ReadFile may not be handled correctly. We recommend loading resources with resources.Get
Changes to Hugo objects (e.g. Page) passed in the template context to lang.Translate may not be detected correctly. We recommend having simple i18n templates without too much data context passed in other than simple types such as strings and numbers.
Note that the cachebuster configuration (when A changes then rebuild B) works well with the above, but we recommend that you revise that configuration, as it in most situations should not be needed. One example where it is still needed is with TailwindCSS and using changes to hugo_stats.json to trigger new CSS rebuilds.
Document Store: Previously, a little simplified, we split the document store (where we store pages and resources) in a tree per language. This worked pretty well, but the structure made some operations harder than they needed to be. We have now restructured it into one Radix tree for all languages. Internally the language is considered to be a dimension of that tree, and the tree can be viewed in all dimensions concurrently. This makes some operations re. language simpler (e.g. finding translations is just a slice range), but the idea is that it should also be relatively inexpensive to add more dimensions if needed (e.g. role).
Fixes #10169
Fixes #10364
Fixes #10482
Fixes #10630
Fixes #10656
Fixes #10694
Fixes #10918
Fixes #11262
Fixes #11439
Fixes #11453
Fixes #11457
Fixes #11466
Fixes #11540
Fixes #11551
Fixes #11556
Fixes #11654
Fixes #11661
Fixes #11663
Fixes #11664
Fixes #11669
Fixes #11671
Fixes #11807
Fixes #11808
Fixes #11809
Fixes #11815
Fixes #11840
Fixes #11853
Fixes #11860
Fixes #11883
Fixes #11904
Fixes #7388
Fixes #7425
Fixes #7436
Fixes #7544
Fixes #7882
Fixes #7960
Fixes #8255
Fixes #8307
Fixes #8863
Fixes #8927
Fixes #9192
Fixes #9324
2023-12-24 13:11:05 -05:00
|
|
|
s.Cfg.LogLevel = logg.LevelError
|
2022-02-09 07:41:04 -05:00
|
|
|
}
|
|
|
|
|
2022-04-16 04:40:21 -04:00
|
|
|
isBinaryRe := regexp.MustCompile(`^(.*)(\.png|\.jpg)$`)
|
|
|
|
|
2023-06-14 02:14:39 -04:00
|
|
|
const dataSourceFilenamePrefix = "sourcefilename:"
|
|
|
|
|
2022-02-09 07:41:04 -05:00
|
|
|
for _, f := range s.data.Files {
|
|
|
|
filename := filepath.Join(s.Cfg.WorkingDir, f.Name)
|
2022-04-16 04:40:21 -04:00
|
|
|
data := bytes.TrimSuffix(f.Data, []byte("\n"))
|
2023-06-14 02:14:39 -04:00
|
|
|
datastr := strings.TrimSpace(string(data))
|
|
|
|
if strings.HasPrefix(datastr, dataSourceFilenamePrefix) {
|
all: Rework page store, add a dynacache, improve partial rebuilds, and some general spring cleaning
There are some breaking changes in this commit, see #11455.
Closes #11455
Closes #11549
This fixes a set of bugs (see issue list) and it is also paying some technical debt accumulated over the years. We now build with Staticcheck enabled in the CI build.
The performance should be about the same as before for regular sized Hugo sites, but it should perform and scale much better to larger data sets, as objects that uses lots of memory (e.g. rendered Markdown, big JSON files read into maps with transform.Unmarshal etc.) will now get automatically garbage collected if needed. Performance on partial rebuilds when running the server in fast render mode should be the same, but the change detection should be much more accurate.
A list of the notable new features:
* A new dependency tracker that covers (almost) all of Hugo's API and is used to do fine grained partial rebuilds when running the server.
* A new and simpler tree document store which allows fast lookups and prefix-walking in all dimensions (e.g. language) concurrently.
* You can now configure an upper memory limit allowing for much larger data sets and/or running on lower specced PCs.
We have lifted the "no resources in sub folders" restriction for branch bundles (e.g. sections).
Memory Limit
* Hugos will, by default, set aside a quarter of the total system memory, but you can set this via the OS environment variable HUGO_MEMORYLIMIT (in gigabytes). This is backed by a partitioned LRU cache used throughout Hugo. A cache that gets dynamically resized in low memory situations, allowing Go's Garbage Collector to free the memory.
New Dependency Tracker: Hugo has had a rule based coarse grained approach to server rebuilds that has worked mostly pretty well, but there have been some surprises (e.g. stale content). This is now revamped with a new dependency tracker that can quickly calculate the delta given a changed resource (e.g. a content file, template, JS file etc.). This handles transitive relations, e.g. $page -> js.Build -> JS import, or $page1.Content -> render hook -> site.GetPage -> $page2.Title, or $page1.Content -> shortcode -> partial -> site.RegularPages -> $page2.Content -> shortcode ..., and should also handle changes to aggregated values (e.g. site.Lastmod) effectively.
This covers all of Hugo's API with 2 known exceptions (a list that may not be fully exhaustive):
Changes to files loaded with template func os.ReadFile may not be handled correctly. We recommend loading resources with resources.Get
Changes to Hugo objects (e.g. Page) passed in the template context to lang.Translate may not be detected correctly. We recommend having simple i18n templates without too much data context passed in other than simple types such as strings and numbers.
Note that the cachebuster configuration (when A changes then rebuild B) works well with the above, but we recommend that you revise that configuration, as it in most situations should not be needed. One example where it is still needed is with TailwindCSS and using changes to hugo_stats.json to trigger new CSS rebuilds.
Document Store: Previously, a little simplified, we split the document store (where we store pages and resources) in a tree per language. This worked pretty well, but the structure made some operations harder than they needed to be. We have now restructured it into one Radix tree for all languages. Internally the language is considered to be a dimension of that tree, and the tree can be viewed in all dimensions concurrently. This makes some operations re. language simpler (e.g. finding translations is just a slice range), but the idea is that it should also be relatively inexpensive to add more dimensions if needed (e.g. role).
Fixes #10169
Fixes #10364
Fixes #10482
Fixes #10630
Fixes #10656
Fixes #10694
Fixes #10918
Fixes #11262
Fixes #11439
Fixes #11453
Fixes #11457
Fixes #11466
Fixes #11540
Fixes #11551
Fixes #11556
Fixes #11654
Fixes #11661
Fixes #11663
Fixes #11664
Fixes #11669
Fixes #11671
Fixes #11807
Fixes #11808
Fixes #11809
Fixes #11815
Fixes #11840
Fixes #11853
Fixes #11860
Fixes #11883
Fixes #11904
Fixes #7388
Fixes #7425
Fixes #7436
Fixes #7544
Fixes #7882
Fixes #7960
Fixes #8255
Fixes #8307
Fixes #8863
Fixes #8927
Fixes #9192
Fixes #9324
2023-12-24 13:11:05 -05:00
|
|
|
// Read from file relative to the current dir.
|
2023-06-14 02:14:39 -04:00
|
|
|
var err error
|
|
|
|
wd, _ := os.Getwd()
|
|
|
|
filename := filepath.Join(wd, strings.TrimSpace(strings.TrimPrefix(datastr, dataSourceFilenamePrefix)))
|
|
|
|
data, err = os.ReadFile(filename)
|
|
|
|
s.Assert(err, qt.IsNil)
|
|
|
|
} else if isBinaryRe.MatchString(filename) {
|
2022-04-16 04:40:21 -04:00
|
|
|
var err error
|
|
|
|
data, err = base64.StdEncoding.DecodeString(string(data))
|
|
|
|
s.Assert(err, qt.IsNil)
|
|
|
|
|
|
|
|
}
|
2023-10-18 12:54:15 -04:00
|
|
|
s.Assert(afs.MkdirAll(filepath.Dir(filename), 0o777), qt.IsNil)
|
|
|
|
s.Assert(afero.WriteFile(afs, filename, data, 0o666), qt.IsNil)
|
2022-02-09 07:41:04 -05:00
|
|
|
}
|
|
|
|
|
2023-01-04 12:24:36 -05:00
|
|
|
configDir := "config"
|
|
|
|
if _, err := afs.Stat(filepath.Join(s.Cfg.WorkingDir, "config")); err != nil {
|
|
|
|
configDir = ""
|
2023-01-15 12:45:51 -05:00
|
|
|
}
|
|
|
|
|
2023-01-04 12:24:36 -05:00
|
|
|
var flags config.Provider
|
|
|
|
if s.Cfg.BaseCfg != nil {
|
|
|
|
flags = s.Cfg.BaseCfg
|
|
|
|
} else {
|
|
|
|
flags = config.New()
|
|
|
|
}
|
|
|
|
|
|
|
|
if s.Cfg.Running {
|
|
|
|
flags.Set("internal", maps.Params{
|
|
|
|
"running": s.Cfg.Running,
|
2023-05-24 03:26:30 -04:00
|
|
|
"watch": s.Cfg.Running,
|
2023-01-04 12:24:36 -05:00
|
|
|
})
|
|
|
|
}
|
|
|
|
|
|
|
|
if s.Cfg.WorkingDir != "" {
|
|
|
|
flags.Set("workingDir", s.Cfg.WorkingDir)
|
|
|
|
}
|
|
|
|
|
all: Rework page store, add a dynacache, improve partial rebuilds, and some general spring cleaning
There are some breaking changes in this commit, see #11455.
Closes #11455
Closes #11549
This fixes a set of bugs (see issue list) and it is also paying some technical debt accumulated over the years. We now build with Staticcheck enabled in the CI build.
The performance should be about the same as before for regular sized Hugo sites, but it should perform and scale much better to larger data sets, as objects that uses lots of memory (e.g. rendered Markdown, big JSON files read into maps with transform.Unmarshal etc.) will now get automatically garbage collected if needed. Performance on partial rebuilds when running the server in fast render mode should be the same, but the change detection should be much more accurate.
A list of the notable new features:
* A new dependency tracker that covers (almost) all of Hugo's API and is used to do fine grained partial rebuilds when running the server.
* A new and simpler tree document store which allows fast lookups and prefix-walking in all dimensions (e.g. language) concurrently.
* You can now configure an upper memory limit allowing for much larger data sets and/or running on lower specced PCs.
We have lifted the "no resources in sub folders" restriction for branch bundles (e.g. sections).
Memory Limit
* Hugos will, by default, set aside a quarter of the total system memory, but you can set this via the OS environment variable HUGO_MEMORYLIMIT (in gigabytes). This is backed by a partitioned LRU cache used throughout Hugo. A cache that gets dynamically resized in low memory situations, allowing Go's Garbage Collector to free the memory.
New Dependency Tracker: Hugo has had a rule based coarse grained approach to server rebuilds that has worked mostly pretty well, but there have been some surprises (e.g. stale content). This is now revamped with a new dependency tracker that can quickly calculate the delta given a changed resource (e.g. a content file, template, JS file etc.). This handles transitive relations, e.g. $page -> js.Build -> JS import, or $page1.Content -> render hook -> site.GetPage -> $page2.Title, or $page1.Content -> shortcode -> partial -> site.RegularPages -> $page2.Content -> shortcode ..., and should also handle changes to aggregated values (e.g. site.Lastmod) effectively.
This covers all of Hugo's API with 2 known exceptions (a list that may not be fully exhaustive):
Changes to files loaded with template func os.ReadFile may not be handled correctly. We recommend loading resources with resources.Get
Changes to Hugo objects (e.g. Page) passed in the template context to lang.Translate may not be detected correctly. We recommend having simple i18n templates without too much data context passed in other than simple types such as strings and numbers.
Note that the cachebuster configuration (when A changes then rebuild B) works well with the above, but we recommend that you revise that configuration, as it in most situations should not be needed. One example where it is still needed is with TailwindCSS and using changes to hugo_stats.json to trigger new CSS rebuilds.
Document Store: Previously, a little simplified, we split the document store (where we store pages and resources) in a tree per language. This worked pretty well, but the structure made some operations harder than they needed to be. We have now restructured it into one Radix tree for all languages. Internally the language is considered to be a dimension of that tree, and the tree can be viewed in all dimensions concurrently. This makes some operations re. language simpler (e.g. finding translations is just a slice range), but the idea is that it should also be relatively inexpensive to add more dimensions if needed (e.g. role).
Fixes #10169
Fixes #10364
Fixes #10482
Fixes #10630
Fixes #10656
Fixes #10694
Fixes #10918
Fixes #11262
Fixes #11439
Fixes #11453
Fixes #11457
Fixes #11466
Fixes #11540
Fixes #11551
Fixes #11556
Fixes #11654
Fixes #11661
Fixes #11663
Fixes #11664
Fixes #11669
Fixes #11671
Fixes #11807
Fixes #11808
Fixes #11809
Fixes #11815
Fixes #11840
Fixes #11853
Fixes #11860
Fixes #11883
Fixes #11904
Fixes #7388
Fixes #7425
Fixes #7436
Fixes #7544
Fixes #7882
Fixes #7960
Fixes #8255
Fixes #8307
Fixes #8863
Fixes #8927
Fixes #9192
Fixes #9324
2023-12-24 13:11:05 -05:00
|
|
|
var w io.Writer
|
|
|
|
if s.Cfg.LogLevel == logg.LevelTrace {
|
|
|
|
w = os.Stdout
|
|
|
|
} else {
|
|
|
|
w = &s.logBuff
|
|
|
|
}
|
2023-07-28 06:04:03 -04:00
|
|
|
|
|
|
|
logger := loggers.New(
|
|
|
|
loggers.Options{
|
2023-11-01 10:15:34 -04:00
|
|
|
Stdout: w,
|
|
|
|
Stderr: w,
|
|
|
|
Level: s.Cfg.LogLevel,
|
|
|
|
DistinctLevel: logg.LevelWarn,
|
2023-07-28 06:04:03 -04:00
|
|
|
},
|
|
|
|
)
|
|
|
|
|
2023-01-04 12:24:36 -05:00
|
|
|
res, err := allconfig.LoadConfig(
|
|
|
|
allconfig.ConfigSourceDescriptor{
|
|
|
|
Flags: flags,
|
|
|
|
ConfigDir: configDir,
|
|
|
|
Fs: afs,
|
2023-07-28 06:04:03 -04:00
|
|
|
Logger: logger,
|
2023-01-04 12:24:36 -05:00
|
|
|
Environ: s.Cfg.Environ,
|
2022-02-09 07:41:04 -05:00
|
|
|
},
|
|
|
|
)
|
2023-01-04 12:24:36 -05:00
|
|
|
if err != nil {
|
|
|
|
initErr = err
|
|
|
|
return
|
|
|
|
}
|
2022-02-09 07:41:04 -05:00
|
|
|
|
2023-01-04 12:24:36 -05:00
|
|
|
fs := hugofs.NewFrom(afs, res.LoadingInfo.BaseConfig)
|
2022-03-21 04:35:15 -04:00
|
|
|
|
|
|
|
s.Assert(err, qt.IsNil)
|
|
|
|
|
2023-07-28 06:04:03 -04:00
|
|
|
depsCfg := deps.DepsCfg{Configs: res, Fs: fs, LogLevel: logger.Level(), LogOut: logger.Out()}
|
2022-02-09 07:41:04 -05:00
|
|
|
sites, err := NewHugoSites(depsCfg)
|
2022-06-04 11:39:34 -04:00
|
|
|
if err != nil {
|
|
|
|
initErr = err
|
|
|
|
return
|
|
|
|
}
|
2023-01-04 12:24:36 -05:00
|
|
|
if sites == nil {
|
|
|
|
initErr = errors.New("no sites")
|
|
|
|
return
|
|
|
|
}
|
2022-02-09 07:41:04 -05:00
|
|
|
|
|
|
|
s.H = sites
|
|
|
|
s.fs = fs
|
|
|
|
|
|
|
|
if s.Cfg.NeedsNpmInstall {
|
|
|
|
wd, _ := os.Getwd()
|
|
|
|
s.Assert(os.Chdir(s.Cfg.WorkingDir), qt.IsNil)
|
|
|
|
s.C.Cleanup(func() { os.Chdir(wd) })
|
|
|
|
sc := security.DefaultConfig
|
2023-06-28 02:56:35 -04:00
|
|
|
sc.Exec.Allow, err = security.NewWhitelist("npm")
|
|
|
|
s.Assert(err, qt.IsNil)
|
2022-02-09 07:41:04 -05:00
|
|
|
ex := hexec.New(sc)
|
|
|
|
command, err := ex.New("npm", "install")
|
|
|
|
s.Assert(err, qt.IsNil)
|
|
|
|
s.Assert(command.Run(), qt.IsNil)
|
|
|
|
|
|
|
|
}
|
|
|
|
})
|
2022-06-04 11:39:34 -04:00
|
|
|
|
|
|
|
return initErr
|
2022-02-09 07:41:04 -05:00
|
|
|
}
|
|
|
|
|
|
|
|
func (s *IntegrationTestBuilder) absFilename(filename string) string {
|
|
|
|
filename = filepath.FromSlash(filename)
|
|
|
|
if filepath.IsAbs(filename) {
|
|
|
|
return filename
|
|
|
|
}
|
|
|
|
if s.Cfg.WorkingDir != "" && !strings.HasPrefix(filename, s.Cfg.WorkingDir) {
|
|
|
|
filename = filepath.Join(s.Cfg.WorkingDir, filename)
|
|
|
|
}
|
|
|
|
return filename
|
|
|
|
}
|
|
|
|
|
all: Rework page store, add a dynacache, improve partial rebuilds, and some general spring cleaning
There are some breaking changes in this commit, see #11455.
Closes #11455
Closes #11549
This fixes a set of bugs (see issue list) and it is also paying some technical debt accumulated over the years. We now build with Staticcheck enabled in the CI build.
The performance should be about the same as before for regular sized Hugo sites, but it should perform and scale much better to larger data sets, as objects that uses lots of memory (e.g. rendered Markdown, big JSON files read into maps with transform.Unmarshal etc.) will now get automatically garbage collected if needed. Performance on partial rebuilds when running the server in fast render mode should be the same, but the change detection should be much more accurate.
A list of the notable new features:
* A new dependency tracker that covers (almost) all of Hugo's API and is used to do fine grained partial rebuilds when running the server.
* A new and simpler tree document store which allows fast lookups and prefix-walking in all dimensions (e.g. language) concurrently.
* You can now configure an upper memory limit allowing for much larger data sets and/or running on lower specced PCs.
We have lifted the "no resources in sub folders" restriction for branch bundles (e.g. sections).
Memory Limit
* Hugos will, by default, set aside a quarter of the total system memory, but you can set this via the OS environment variable HUGO_MEMORYLIMIT (in gigabytes). This is backed by a partitioned LRU cache used throughout Hugo. A cache that gets dynamically resized in low memory situations, allowing Go's Garbage Collector to free the memory.
New Dependency Tracker: Hugo has had a rule based coarse grained approach to server rebuilds that has worked mostly pretty well, but there have been some surprises (e.g. stale content). This is now revamped with a new dependency tracker that can quickly calculate the delta given a changed resource (e.g. a content file, template, JS file etc.). This handles transitive relations, e.g. $page -> js.Build -> JS import, or $page1.Content -> render hook -> site.GetPage -> $page2.Title, or $page1.Content -> shortcode -> partial -> site.RegularPages -> $page2.Content -> shortcode ..., and should also handle changes to aggregated values (e.g. site.Lastmod) effectively.
This covers all of Hugo's API with 2 known exceptions (a list that may not be fully exhaustive):
Changes to files loaded with template func os.ReadFile may not be handled correctly. We recommend loading resources with resources.Get
Changes to Hugo objects (e.g. Page) passed in the template context to lang.Translate may not be detected correctly. We recommend having simple i18n templates without too much data context passed in other than simple types such as strings and numbers.
Note that the cachebuster configuration (when A changes then rebuild B) works well with the above, but we recommend that you revise that configuration, as it in most situations should not be needed. One example where it is still needed is with TailwindCSS and using changes to hugo_stats.json to trigger new CSS rebuilds.
Document Store: Previously, a little simplified, we split the document store (where we store pages and resources) in a tree per language. This worked pretty well, but the structure made some operations harder than they needed to be. We have now restructured it into one Radix tree for all languages. Internally the language is considered to be a dimension of that tree, and the tree can be viewed in all dimensions concurrently. This makes some operations re. language simpler (e.g. finding translations is just a slice range), but the idea is that it should also be relatively inexpensive to add more dimensions if needed (e.g. role).
Fixes #10169
Fixes #10364
Fixes #10482
Fixes #10630
Fixes #10656
Fixes #10694
Fixes #10918
Fixes #11262
Fixes #11439
Fixes #11453
Fixes #11457
Fixes #11466
Fixes #11540
Fixes #11551
Fixes #11556
Fixes #11654
Fixes #11661
Fixes #11663
Fixes #11664
Fixes #11669
Fixes #11671
Fixes #11807
Fixes #11808
Fixes #11809
Fixes #11815
Fixes #11840
Fixes #11853
Fixes #11860
Fixes #11883
Fixes #11904
Fixes #7388
Fixes #7425
Fixes #7436
Fixes #7544
Fixes #7882
Fixes #7960
Fixes #8255
Fixes #8307
Fixes #8863
Fixes #8927
Fixes #9192
Fixes #9324
2023-12-24 13:11:05 -05:00
|
|
|
func (s *IntegrationTestBuilder) reset() {
|
|
|
|
s.changedFiles = nil
|
|
|
|
s.createdFiles = nil
|
|
|
|
s.removedFiles = nil
|
|
|
|
s.renamedFiles = nil
|
|
|
|
}
|
|
|
|
|
2022-02-09 07:41:04 -05:00
|
|
|
func (s *IntegrationTestBuilder) build(cfg BuildCfg) error {
|
|
|
|
s.Helper()
|
|
|
|
defer func() {
|
all: Rework page store, add a dynacache, improve partial rebuilds, and some general spring cleaning
There are some breaking changes in this commit, see #11455.
Closes #11455
Closes #11549
This fixes a set of bugs (see issue list) and it is also paying some technical debt accumulated over the years. We now build with Staticcheck enabled in the CI build.
The performance should be about the same as before for regular sized Hugo sites, but it should perform and scale much better to larger data sets, as objects that uses lots of memory (e.g. rendered Markdown, big JSON files read into maps with transform.Unmarshal etc.) will now get automatically garbage collected if needed. Performance on partial rebuilds when running the server in fast render mode should be the same, but the change detection should be much more accurate.
A list of the notable new features:
* A new dependency tracker that covers (almost) all of Hugo's API and is used to do fine grained partial rebuilds when running the server.
* A new and simpler tree document store which allows fast lookups and prefix-walking in all dimensions (e.g. language) concurrently.
* You can now configure an upper memory limit allowing for much larger data sets and/or running on lower specced PCs.
We have lifted the "no resources in sub folders" restriction for branch bundles (e.g. sections).
Memory Limit
* Hugos will, by default, set aside a quarter of the total system memory, but you can set this via the OS environment variable HUGO_MEMORYLIMIT (in gigabytes). This is backed by a partitioned LRU cache used throughout Hugo. A cache that gets dynamically resized in low memory situations, allowing Go's Garbage Collector to free the memory.
New Dependency Tracker: Hugo has had a rule based coarse grained approach to server rebuilds that has worked mostly pretty well, but there have been some surprises (e.g. stale content). This is now revamped with a new dependency tracker that can quickly calculate the delta given a changed resource (e.g. a content file, template, JS file etc.). This handles transitive relations, e.g. $page -> js.Build -> JS import, or $page1.Content -> render hook -> site.GetPage -> $page2.Title, or $page1.Content -> shortcode -> partial -> site.RegularPages -> $page2.Content -> shortcode ..., and should also handle changes to aggregated values (e.g. site.Lastmod) effectively.
This covers all of Hugo's API with 2 known exceptions (a list that may not be fully exhaustive):
Changes to files loaded with template func os.ReadFile may not be handled correctly. We recommend loading resources with resources.Get
Changes to Hugo objects (e.g. Page) passed in the template context to lang.Translate may not be detected correctly. We recommend having simple i18n templates without too much data context passed in other than simple types such as strings and numbers.
Note that the cachebuster configuration (when A changes then rebuild B) works well with the above, but we recommend that you revise that configuration, as it in most situations should not be needed. One example where it is still needed is with TailwindCSS and using changes to hugo_stats.json to trigger new CSS rebuilds.
Document Store: Previously, a little simplified, we split the document store (where we store pages and resources) in a tree per language. This worked pretty well, but the structure made some operations harder than they needed to be. We have now restructured it into one Radix tree for all languages. Internally the language is considered to be a dimension of that tree, and the tree can be viewed in all dimensions concurrently. This makes some operations re. language simpler (e.g. finding translations is just a slice range), but the idea is that it should also be relatively inexpensive to add more dimensions if needed (e.g. role).
Fixes #10169
Fixes #10364
Fixes #10482
Fixes #10630
Fixes #10656
Fixes #10694
Fixes #10918
Fixes #11262
Fixes #11439
Fixes #11453
Fixes #11457
Fixes #11466
Fixes #11540
Fixes #11551
Fixes #11556
Fixes #11654
Fixes #11661
Fixes #11663
Fixes #11664
Fixes #11669
Fixes #11671
Fixes #11807
Fixes #11808
Fixes #11809
Fixes #11815
Fixes #11840
Fixes #11853
Fixes #11860
Fixes #11883
Fixes #11904
Fixes #7388
Fixes #7425
Fixes #7436
Fixes #7544
Fixes #7882
Fixes #7960
Fixes #8255
Fixes #8307
Fixes #8863
Fixes #8927
Fixes #9192
Fixes #9324
2023-12-24 13:11:05 -05:00
|
|
|
s.reset()
|
2022-02-09 07:41:04 -05:00
|
|
|
}()
|
|
|
|
|
|
|
|
changeEvents := s.changeEvents()
|
|
|
|
s.logBuff.Reset()
|
all: Rework page store, add a dynacache, improve partial rebuilds, and some general spring cleaning
There are some breaking changes in this commit, see #11455.
Closes #11455
Closes #11549
This fixes a set of bugs (see issue list) and it is also paying some technical debt accumulated over the years. We now build with Staticcheck enabled in the CI build.
The performance should be about the same as before for regular sized Hugo sites, but it should perform and scale much better to larger data sets, as objects that uses lots of memory (e.g. rendered Markdown, big JSON files read into maps with transform.Unmarshal etc.) will now get automatically garbage collected if needed. Performance on partial rebuilds when running the server in fast render mode should be the same, but the change detection should be much more accurate.
A list of the notable new features:
* A new dependency tracker that covers (almost) all of Hugo's API and is used to do fine grained partial rebuilds when running the server.
* A new and simpler tree document store which allows fast lookups and prefix-walking in all dimensions (e.g. language) concurrently.
* You can now configure an upper memory limit allowing for much larger data sets and/or running on lower specced PCs.
We have lifted the "no resources in sub folders" restriction for branch bundles (e.g. sections).
Memory Limit
* Hugos will, by default, set aside a quarter of the total system memory, but you can set this via the OS environment variable HUGO_MEMORYLIMIT (in gigabytes). This is backed by a partitioned LRU cache used throughout Hugo. A cache that gets dynamically resized in low memory situations, allowing Go's Garbage Collector to free the memory.
New Dependency Tracker: Hugo has had a rule based coarse grained approach to server rebuilds that has worked mostly pretty well, but there have been some surprises (e.g. stale content). This is now revamped with a new dependency tracker that can quickly calculate the delta given a changed resource (e.g. a content file, template, JS file etc.). This handles transitive relations, e.g. $page -> js.Build -> JS import, or $page1.Content -> render hook -> site.GetPage -> $page2.Title, or $page1.Content -> shortcode -> partial -> site.RegularPages -> $page2.Content -> shortcode ..., and should also handle changes to aggregated values (e.g. site.Lastmod) effectively.
This covers all of Hugo's API with 2 known exceptions (a list that may not be fully exhaustive):
Changes to files loaded with template func os.ReadFile may not be handled correctly. We recommend loading resources with resources.Get
Changes to Hugo objects (e.g. Page) passed in the template context to lang.Translate may not be detected correctly. We recommend having simple i18n templates without too much data context passed in other than simple types such as strings and numbers.
Note that the cachebuster configuration (when A changes then rebuild B) works well with the above, but we recommend that you revise that configuration, as it in most situations should not be needed. One example where it is still needed is with TailwindCSS and using changes to hugo_stats.json to trigger new CSS rebuilds.
Document Store: Previously, a little simplified, we split the document store (where we store pages and resources) in a tree per language. This worked pretty well, but the structure made some operations harder than they needed to be. We have now restructured it into one Radix tree for all languages. Internally the language is considered to be a dimension of that tree, and the tree can be viewed in all dimensions concurrently. This makes some operations re. language simpler (e.g. finding translations is just a slice range), but the idea is that it should also be relatively inexpensive to add more dimensions if needed (e.g. role).
Fixes #10169
Fixes #10364
Fixes #10482
Fixes #10630
Fixes #10656
Fixes #10694
Fixes #10918
Fixes #11262
Fixes #11439
Fixes #11453
Fixes #11457
Fixes #11466
Fixes #11540
Fixes #11551
Fixes #11556
Fixes #11654
Fixes #11661
Fixes #11663
Fixes #11664
Fixes #11669
Fixes #11671
Fixes #11807
Fixes #11808
Fixes #11809
Fixes #11815
Fixes #11840
Fixes #11853
Fixes #11860
Fixes #11883
Fixes #11904
Fixes #7388
Fixes #7425
Fixes #7436
Fixes #7544
Fixes #7882
Fixes #7960
Fixes #8255
Fixes #8307
Fixes #8863
Fixes #8927
Fixes #9192
Fixes #9324
2023-12-24 13:11:05 -05:00
|
|
|
s.counters = &buildCounters{}
|
2022-02-09 07:41:04 -05:00
|
|
|
cfg.testCounters = s.counters
|
|
|
|
|
|
|
|
if s.buildCount > 0 && (len(changeEvents) == 0) {
|
|
|
|
return nil
|
|
|
|
}
|
|
|
|
|
|
|
|
s.buildCount++
|
|
|
|
|
|
|
|
err := s.H.Build(cfg, changeEvents...)
|
|
|
|
if err != nil {
|
|
|
|
return err
|
|
|
|
}
|
|
|
|
logErrorCount := s.H.NumLogErrors()
|
|
|
|
if logErrorCount > 0 {
|
|
|
|
return fmt.Errorf("logged %d error(s): %s", logErrorCount, s.logBuff.String())
|
|
|
|
}
|
|
|
|
|
|
|
|
return nil
|
|
|
|
}
|
|
|
|
|
|
|
|
func (s *IntegrationTestBuilder) changeEvents() []fsnotify.Event {
|
|
|
|
var events []fsnotify.Event
|
|
|
|
for _, v := range s.removedFiles {
|
|
|
|
events = append(events, fsnotify.Event{
|
|
|
|
Name: v,
|
|
|
|
Op: fsnotify.Remove,
|
|
|
|
})
|
|
|
|
}
|
|
|
|
for _, v := range s.renamedFiles {
|
|
|
|
events = append(events, fsnotify.Event{
|
|
|
|
Name: v,
|
|
|
|
Op: fsnotify.Rename,
|
|
|
|
})
|
|
|
|
}
|
all: Rework page store, add a dynacache, improve partial rebuilds, and some general spring cleaning
There are some breaking changes in this commit, see #11455.
Closes #11455
Closes #11549
This fixes a set of bugs (see issue list) and it is also paying some technical debt accumulated over the years. We now build with Staticcheck enabled in the CI build.
The performance should be about the same as before for regular sized Hugo sites, but it should perform and scale much better to larger data sets, as objects that uses lots of memory (e.g. rendered Markdown, big JSON files read into maps with transform.Unmarshal etc.) will now get automatically garbage collected if needed. Performance on partial rebuilds when running the server in fast render mode should be the same, but the change detection should be much more accurate.
A list of the notable new features:
* A new dependency tracker that covers (almost) all of Hugo's API and is used to do fine grained partial rebuilds when running the server.
* A new and simpler tree document store which allows fast lookups and prefix-walking in all dimensions (e.g. language) concurrently.
* You can now configure an upper memory limit allowing for much larger data sets and/or running on lower specced PCs.
We have lifted the "no resources in sub folders" restriction for branch bundles (e.g. sections).
Memory Limit
* Hugos will, by default, set aside a quarter of the total system memory, but you can set this via the OS environment variable HUGO_MEMORYLIMIT (in gigabytes). This is backed by a partitioned LRU cache used throughout Hugo. A cache that gets dynamically resized in low memory situations, allowing Go's Garbage Collector to free the memory.
New Dependency Tracker: Hugo has had a rule based coarse grained approach to server rebuilds that has worked mostly pretty well, but there have been some surprises (e.g. stale content). This is now revamped with a new dependency tracker that can quickly calculate the delta given a changed resource (e.g. a content file, template, JS file etc.). This handles transitive relations, e.g. $page -> js.Build -> JS import, or $page1.Content -> render hook -> site.GetPage -> $page2.Title, or $page1.Content -> shortcode -> partial -> site.RegularPages -> $page2.Content -> shortcode ..., and should also handle changes to aggregated values (e.g. site.Lastmod) effectively.
This covers all of Hugo's API with 2 known exceptions (a list that may not be fully exhaustive):
Changes to files loaded with template func os.ReadFile may not be handled correctly. We recommend loading resources with resources.Get
Changes to Hugo objects (e.g. Page) passed in the template context to lang.Translate may not be detected correctly. We recommend having simple i18n templates without too much data context passed in other than simple types such as strings and numbers.
Note that the cachebuster configuration (when A changes then rebuild B) works well with the above, but we recommend that you revise that configuration, as it in most situations should not be needed. One example where it is still needed is with TailwindCSS and using changes to hugo_stats.json to trigger new CSS rebuilds.
Document Store: Previously, a little simplified, we split the document store (where we store pages and resources) in a tree per language. This worked pretty well, but the structure made some operations harder than they needed to be. We have now restructured it into one Radix tree for all languages. Internally the language is considered to be a dimension of that tree, and the tree can be viewed in all dimensions concurrently. This makes some operations re. language simpler (e.g. finding translations is just a slice range), but the idea is that it should also be relatively inexpensive to add more dimensions if needed (e.g. role).
Fixes #10169
Fixes #10364
Fixes #10482
Fixes #10630
Fixes #10656
Fixes #10694
Fixes #10918
Fixes #11262
Fixes #11439
Fixes #11453
Fixes #11457
Fixes #11466
Fixes #11540
Fixes #11551
Fixes #11556
Fixes #11654
Fixes #11661
Fixes #11663
Fixes #11664
Fixes #11669
Fixes #11671
Fixes #11807
Fixes #11808
Fixes #11809
Fixes #11815
Fixes #11840
Fixes #11853
Fixes #11860
Fixes #11883
Fixes #11904
Fixes #7388
Fixes #7425
Fixes #7436
Fixes #7544
Fixes #7882
Fixes #7960
Fixes #8255
Fixes #8307
Fixes #8863
Fixes #8927
Fixes #9192
Fixes #9324
2023-12-24 13:11:05 -05:00
|
|
|
|
|
|
|
for _, v := range s.renamedDirs {
|
|
|
|
events = append(events, fsnotify.Event{
|
|
|
|
Name: v,
|
|
|
|
// This is what we get on MacOS.
|
|
|
|
Op: fsnotify.Remove | fsnotify.Rename,
|
|
|
|
})
|
|
|
|
}
|
|
|
|
|
2022-02-09 07:41:04 -05:00
|
|
|
for _, v := range s.changedFiles {
|
|
|
|
events = append(events, fsnotify.Event{
|
|
|
|
Name: v,
|
|
|
|
Op: fsnotify.Write,
|
|
|
|
})
|
|
|
|
}
|
|
|
|
for _, v := range s.createdFiles {
|
|
|
|
events = append(events, fsnotify.Event{
|
|
|
|
Name: v,
|
|
|
|
Op: fsnotify.Create,
|
|
|
|
})
|
|
|
|
}
|
|
|
|
|
all: Rework page store, add a dynacache, improve partial rebuilds, and some general spring cleaning
There are some breaking changes in this commit, see #11455.
Closes #11455
Closes #11549
This fixes a set of bugs (see issue list) and it is also paying some technical debt accumulated over the years. We now build with Staticcheck enabled in the CI build.
The performance should be about the same as before for regular sized Hugo sites, but it should perform and scale much better to larger data sets, as objects that uses lots of memory (e.g. rendered Markdown, big JSON files read into maps with transform.Unmarshal etc.) will now get automatically garbage collected if needed. Performance on partial rebuilds when running the server in fast render mode should be the same, but the change detection should be much more accurate.
A list of the notable new features:
* A new dependency tracker that covers (almost) all of Hugo's API and is used to do fine grained partial rebuilds when running the server.
* A new and simpler tree document store which allows fast lookups and prefix-walking in all dimensions (e.g. language) concurrently.
* You can now configure an upper memory limit allowing for much larger data sets and/or running on lower specced PCs.
We have lifted the "no resources in sub folders" restriction for branch bundles (e.g. sections).
Memory Limit
* Hugos will, by default, set aside a quarter of the total system memory, but you can set this via the OS environment variable HUGO_MEMORYLIMIT (in gigabytes). This is backed by a partitioned LRU cache used throughout Hugo. A cache that gets dynamically resized in low memory situations, allowing Go's Garbage Collector to free the memory.
New Dependency Tracker: Hugo has had a rule based coarse grained approach to server rebuilds that has worked mostly pretty well, but there have been some surprises (e.g. stale content). This is now revamped with a new dependency tracker that can quickly calculate the delta given a changed resource (e.g. a content file, template, JS file etc.). This handles transitive relations, e.g. $page -> js.Build -> JS import, or $page1.Content -> render hook -> site.GetPage -> $page2.Title, or $page1.Content -> shortcode -> partial -> site.RegularPages -> $page2.Content -> shortcode ..., and should also handle changes to aggregated values (e.g. site.Lastmod) effectively.
This covers all of Hugo's API with 2 known exceptions (a list that may not be fully exhaustive):
Changes to files loaded with template func os.ReadFile may not be handled correctly. We recommend loading resources with resources.Get
Changes to Hugo objects (e.g. Page) passed in the template context to lang.Translate may not be detected correctly. We recommend having simple i18n templates without too much data context passed in other than simple types such as strings and numbers.
Note that the cachebuster configuration (when A changes then rebuild B) works well with the above, but we recommend that you revise that configuration, as it in most situations should not be needed. One example where it is still needed is with TailwindCSS and using changes to hugo_stats.json to trigger new CSS rebuilds.
Document Store: Previously, a little simplified, we split the document store (where we store pages and resources) in a tree per language. This worked pretty well, but the structure made some operations harder than they needed to be. We have now restructured it into one Radix tree for all languages. Internally the language is considered to be a dimension of that tree, and the tree can be viewed in all dimensions concurrently. This makes some operations re. language simpler (e.g. finding translations is just a slice range), but the idea is that it should also be relatively inexpensive to add more dimensions if needed (e.g. role).
Fixes #10169
Fixes #10364
Fixes #10482
Fixes #10630
Fixes #10656
Fixes #10694
Fixes #10918
Fixes #11262
Fixes #11439
Fixes #11453
Fixes #11457
Fixes #11466
Fixes #11540
Fixes #11551
Fixes #11556
Fixes #11654
Fixes #11661
Fixes #11663
Fixes #11664
Fixes #11669
Fixes #11671
Fixes #11807
Fixes #11808
Fixes #11809
Fixes #11815
Fixes #11840
Fixes #11853
Fixes #11860
Fixes #11883
Fixes #11904
Fixes #7388
Fixes #7425
Fixes #7436
Fixes #7544
Fixes #7882
Fixes #7960
Fixes #8255
Fixes #8307
Fixes #8863
Fixes #8927
Fixes #9192
Fixes #9324
2023-12-24 13:11:05 -05:00
|
|
|
// Shuffle events.
|
|
|
|
for i := range events {
|
|
|
|
j := rand.Intn(i + 1)
|
|
|
|
events[i], events[j] = events[j], events[i]
|
|
|
|
}
|
|
|
|
|
2022-02-09 07:41:04 -05:00
|
|
|
return events
|
|
|
|
}
|
|
|
|
|
2022-03-21 04:35:15 -04:00
|
|
|
func (s *IntegrationTestBuilder) readWorkingDir(t testing.TB, fs *hugofs.Fs, filename string) string {
|
2022-02-09 07:41:04 -05:00
|
|
|
t.Helper()
|
2022-03-21 04:35:15 -04:00
|
|
|
return s.readFileFromFs(t, fs.WorkingDirReadOnly, filename)
|
2022-02-09 07:41:04 -05:00
|
|
|
}
|
|
|
|
|
|
|
|
func (s *IntegrationTestBuilder) readFileFromFs(t testing.TB, fs afero.Fs, filename string) string {
|
|
|
|
t.Helper()
|
|
|
|
filename = filepath.Clean(filename)
|
|
|
|
b, err := afero.ReadFile(fs, filename)
|
|
|
|
if err != nil {
|
|
|
|
// Print some debug info
|
|
|
|
hadSlash := strings.HasPrefix(filename, helpers.FilePathSeparator)
|
|
|
|
start := 0
|
|
|
|
if hadSlash {
|
|
|
|
start = 1
|
|
|
|
}
|
|
|
|
end := start + 1
|
|
|
|
|
|
|
|
parts := strings.Split(filename, helpers.FilePathSeparator)
|
|
|
|
if parts[start] == "work" {
|
|
|
|
end++
|
|
|
|
}
|
|
|
|
|
|
|
|
s.Assert(err, qt.IsNil)
|
|
|
|
|
|
|
|
}
|
|
|
|
return string(b)
|
|
|
|
}
|
|
|
|
|
|
|
|
func (s *IntegrationTestBuilder) writeSource(filename, content string) {
|
|
|
|
s.Helper()
|
|
|
|
s.writeToFs(s.fs.Source, filename, content)
|
|
|
|
}
|
|
|
|
|
|
|
|
func (s *IntegrationTestBuilder) writeToFs(fs afero.Fs, filename, content string) {
|
|
|
|
s.Helper()
|
2023-10-18 12:54:15 -04:00
|
|
|
if err := afero.WriteFile(fs, filepath.FromSlash(filename), []byte(content), 0o755); err != nil {
|
2022-02-09 07:41:04 -05:00
|
|
|
s.Fatalf("Failed to write file: %s", err)
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
type IntegrationTestConfig struct {
|
|
|
|
T testing.TB
|
|
|
|
|
|
|
|
// The files to use on txtar format, see
|
|
|
|
// https://pkg.go.dev/golang.org/x/exp/cmd/txtar
|
|
|
|
TxtarString string
|
|
|
|
|
2023-01-04 12:24:36 -05:00
|
|
|
// COnfig to use as the base. We will also read the config from the txtar.
|
|
|
|
BaseCfg config.Provider
|
|
|
|
|
|
|
|
// Environment variables passed to the config loader.
|
|
|
|
Environ []string
|
|
|
|
|
2022-02-09 07:41:04 -05:00
|
|
|
// Whether to simulate server mode.
|
|
|
|
Running bool
|
|
|
|
|
|
|
|
// Will print the log buffer after the build
|
|
|
|
Verbose bool
|
|
|
|
|
all: Rework page store, add a dynacache, improve partial rebuilds, and some general spring cleaning
There are some breaking changes in this commit, see #11455.
Closes #11455
Closes #11549
This fixes a set of bugs (see issue list) and it is also paying some technical debt accumulated over the years. We now build with Staticcheck enabled in the CI build.
The performance should be about the same as before for regular sized Hugo sites, but it should perform and scale much better to larger data sets, as objects that uses lots of memory (e.g. rendered Markdown, big JSON files read into maps with transform.Unmarshal etc.) will now get automatically garbage collected if needed. Performance on partial rebuilds when running the server in fast render mode should be the same, but the change detection should be much more accurate.
A list of the notable new features:
* A new dependency tracker that covers (almost) all of Hugo's API and is used to do fine grained partial rebuilds when running the server.
* A new and simpler tree document store which allows fast lookups and prefix-walking in all dimensions (e.g. language) concurrently.
* You can now configure an upper memory limit allowing for much larger data sets and/or running on lower specced PCs.
We have lifted the "no resources in sub folders" restriction for branch bundles (e.g. sections).
Memory Limit
* Hugos will, by default, set aside a quarter of the total system memory, but you can set this via the OS environment variable HUGO_MEMORYLIMIT (in gigabytes). This is backed by a partitioned LRU cache used throughout Hugo. A cache that gets dynamically resized in low memory situations, allowing Go's Garbage Collector to free the memory.
New Dependency Tracker: Hugo has had a rule based coarse grained approach to server rebuilds that has worked mostly pretty well, but there have been some surprises (e.g. stale content). This is now revamped with a new dependency tracker that can quickly calculate the delta given a changed resource (e.g. a content file, template, JS file etc.). This handles transitive relations, e.g. $page -> js.Build -> JS import, or $page1.Content -> render hook -> site.GetPage -> $page2.Title, or $page1.Content -> shortcode -> partial -> site.RegularPages -> $page2.Content -> shortcode ..., and should also handle changes to aggregated values (e.g. site.Lastmod) effectively.
This covers all of Hugo's API with 2 known exceptions (a list that may not be fully exhaustive):
Changes to files loaded with template func os.ReadFile may not be handled correctly. We recommend loading resources with resources.Get
Changes to Hugo objects (e.g. Page) passed in the template context to lang.Translate may not be detected correctly. We recommend having simple i18n templates without too much data context passed in other than simple types such as strings and numbers.
Note that the cachebuster configuration (when A changes then rebuild B) works well with the above, but we recommend that you revise that configuration, as it in most situations should not be needed. One example where it is still needed is with TailwindCSS and using changes to hugo_stats.json to trigger new CSS rebuilds.
Document Store: Previously, a little simplified, we split the document store (where we store pages and resources) in a tree per language. This worked pretty well, but the structure made some operations harder than they needed to be. We have now restructured it into one Radix tree for all languages. Internally the language is considered to be a dimension of that tree, and the tree can be viewed in all dimensions concurrently. This makes some operations re. language simpler (e.g. finding translations is just a slice range), but the idea is that it should also be relatively inexpensive to add more dimensions if needed (e.g. role).
Fixes #10169
Fixes #10364
Fixes #10482
Fixes #10630
Fixes #10656
Fixes #10694
Fixes #10918
Fixes #11262
Fixes #11439
Fixes #11453
Fixes #11457
Fixes #11466
Fixes #11540
Fixes #11551
Fixes #11556
Fixes #11654
Fixes #11661
Fixes #11663
Fixes #11664
Fixes #11669
Fixes #11671
Fixes #11807
Fixes #11808
Fixes #11809
Fixes #11815
Fixes #11840
Fixes #11853
Fixes #11860
Fixes #11883
Fixes #11904
Fixes #7388
Fixes #7425
Fixes #7436
Fixes #7544
Fixes #7882
Fixes #7960
Fixes #8255
Fixes #8307
Fixes #8863
Fixes #8927
Fixes #9192
Fixes #9324
2023-12-24 13:11:05 -05:00
|
|
|
// The log level to use.
|
2023-06-16 02:17:42 -04:00
|
|
|
LogLevel logg.Level
|
2022-02-09 07:41:04 -05:00
|
|
|
|
|
|
|
// Whether it needs the real file system (e.g. for js.Build tests).
|
|
|
|
NeedsOsFS bool
|
|
|
|
|
2023-03-04 05:09:03 -05:00
|
|
|
// Whether to run GC after each build.
|
|
|
|
RunGC bool
|
|
|
|
|
2022-04-16 04:40:21 -04:00
|
|
|
// Do not remove the temp dir after the test.
|
|
|
|
PrintAndKeepTempDir bool
|
|
|
|
|
2022-02-09 07:41:04 -05:00
|
|
|
// Whether to run npm install before Build.
|
|
|
|
NeedsNpmInstall bool
|
|
|
|
|
all: Rework page store, add a dynacache, improve partial rebuilds, and some general spring cleaning
There are some breaking changes in this commit, see #11455.
Closes #11455
Closes #11549
This fixes a set of bugs (see issue list) and it is also paying some technical debt accumulated over the years. We now build with Staticcheck enabled in the CI build.
The performance should be about the same as before for regular sized Hugo sites, but it should perform and scale much better to larger data sets, as objects that uses lots of memory (e.g. rendered Markdown, big JSON files read into maps with transform.Unmarshal etc.) will now get automatically garbage collected if needed. Performance on partial rebuilds when running the server in fast render mode should be the same, but the change detection should be much more accurate.
A list of the notable new features:
* A new dependency tracker that covers (almost) all of Hugo's API and is used to do fine grained partial rebuilds when running the server.
* A new and simpler tree document store which allows fast lookups and prefix-walking in all dimensions (e.g. language) concurrently.
* You can now configure an upper memory limit allowing for much larger data sets and/or running on lower specced PCs.
We have lifted the "no resources in sub folders" restriction for branch bundles (e.g. sections).
Memory Limit
* Hugos will, by default, set aside a quarter of the total system memory, but you can set this via the OS environment variable HUGO_MEMORYLIMIT (in gigabytes). This is backed by a partitioned LRU cache used throughout Hugo. A cache that gets dynamically resized in low memory situations, allowing Go's Garbage Collector to free the memory.
New Dependency Tracker: Hugo has had a rule based coarse grained approach to server rebuilds that has worked mostly pretty well, but there have been some surprises (e.g. stale content). This is now revamped with a new dependency tracker that can quickly calculate the delta given a changed resource (e.g. a content file, template, JS file etc.). This handles transitive relations, e.g. $page -> js.Build -> JS import, or $page1.Content -> render hook -> site.GetPage -> $page2.Title, or $page1.Content -> shortcode -> partial -> site.RegularPages -> $page2.Content -> shortcode ..., and should also handle changes to aggregated values (e.g. site.Lastmod) effectively.
This covers all of Hugo's API with 2 known exceptions (a list that may not be fully exhaustive):
Changes to files loaded with template func os.ReadFile may not be handled correctly. We recommend loading resources with resources.Get
Changes to Hugo objects (e.g. Page) passed in the template context to lang.Translate may not be detected correctly. We recommend having simple i18n templates without too much data context passed in other than simple types such as strings and numbers.
Note that the cachebuster configuration (when A changes then rebuild B) works well with the above, but we recommend that you revise that configuration, as it in most situations should not be needed. One example where it is still needed is with TailwindCSS and using changes to hugo_stats.json to trigger new CSS rebuilds.
Document Store: Previously, a little simplified, we split the document store (where we store pages and resources) in a tree per language. This worked pretty well, but the structure made some operations harder than they needed to be. We have now restructured it into one Radix tree for all languages. Internally the language is considered to be a dimension of that tree, and the tree can be viewed in all dimensions concurrently. This makes some operations re. language simpler (e.g. finding translations is just a slice range), but the idea is that it should also be relatively inexpensive to add more dimensions if needed (e.g. role).
Fixes #10169
Fixes #10364
Fixes #10482
Fixes #10630
Fixes #10656
Fixes #10694
Fixes #10918
Fixes #11262
Fixes #11439
Fixes #11453
Fixes #11457
Fixes #11466
Fixes #11540
Fixes #11551
Fixes #11556
Fixes #11654
Fixes #11661
Fixes #11663
Fixes #11664
Fixes #11669
Fixes #11671
Fixes #11807
Fixes #11808
Fixes #11809
Fixes #11815
Fixes #11840
Fixes #11853
Fixes #11860
Fixes #11883
Fixes #11904
Fixes #7388
Fixes #7425
Fixes #7436
Fixes #7544
Fixes #7882
Fixes #7960
Fixes #8255
Fixes #8307
Fixes #8863
Fixes #8927
Fixes #9192
Fixes #9324
2023-12-24 13:11:05 -05:00
|
|
|
// Whether to normalize the Unicode filenames to NFD on Darwin.
|
|
|
|
NFDFormOnDarwin bool
|
|
|
|
|
|
|
|
// The working dir to use. If not absolute, a temp dir will be created.
|
2022-02-09 07:41:04 -05:00
|
|
|
WorkingDir string
|
2023-05-15 02:39:52 -04:00
|
|
|
|
all: Rework page store, add a dynacache, improve partial rebuilds, and some general spring cleaning
There are some breaking changes in this commit, see #11455.
Closes #11455
Closes #11549
This fixes a set of bugs (see issue list) and it is also paying some technical debt accumulated over the years. We now build with Staticcheck enabled in the CI build.
The performance should be about the same as before for regular sized Hugo sites, but it should perform and scale much better to larger data sets, as objects that uses lots of memory (e.g. rendered Markdown, big JSON files read into maps with transform.Unmarshal etc.) will now get automatically garbage collected if needed. Performance on partial rebuilds when running the server in fast render mode should be the same, but the change detection should be much more accurate.
A list of the notable new features:
* A new dependency tracker that covers (almost) all of Hugo's API and is used to do fine grained partial rebuilds when running the server.
* A new and simpler tree document store which allows fast lookups and prefix-walking in all dimensions (e.g. language) concurrently.
* You can now configure an upper memory limit allowing for much larger data sets and/or running on lower specced PCs.
We have lifted the "no resources in sub folders" restriction for branch bundles (e.g. sections).
Memory Limit
* Hugos will, by default, set aside a quarter of the total system memory, but you can set this via the OS environment variable HUGO_MEMORYLIMIT (in gigabytes). This is backed by a partitioned LRU cache used throughout Hugo. A cache that gets dynamically resized in low memory situations, allowing Go's Garbage Collector to free the memory.
New Dependency Tracker: Hugo has had a rule based coarse grained approach to server rebuilds that has worked mostly pretty well, but there have been some surprises (e.g. stale content). This is now revamped with a new dependency tracker that can quickly calculate the delta given a changed resource (e.g. a content file, template, JS file etc.). This handles transitive relations, e.g. $page -> js.Build -> JS import, or $page1.Content -> render hook -> site.GetPage -> $page2.Title, or $page1.Content -> shortcode -> partial -> site.RegularPages -> $page2.Content -> shortcode ..., and should also handle changes to aggregated values (e.g. site.Lastmod) effectively.
This covers all of Hugo's API with 2 known exceptions (a list that may not be fully exhaustive):
Changes to files loaded with template func os.ReadFile may not be handled correctly. We recommend loading resources with resources.Get
Changes to Hugo objects (e.g. Page) passed in the template context to lang.Translate may not be detected correctly. We recommend having simple i18n templates without too much data context passed in other than simple types such as strings and numbers.
Note that the cachebuster configuration (when A changes then rebuild B) works well with the above, but we recommend that you revise that configuration, as it in most situations should not be needed. One example where it is still needed is with TailwindCSS and using changes to hugo_stats.json to trigger new CSS rebuilds.
Document Store: Previously, a little simplified, we split the document store (where we store pages and resources) in a tree per language. This worked pretty well, but the structure made some operations harder than they needed to be. We have now restructured it into one Radix tree for all languages. Internally the language is considered to be a dimension of that tree, and the tree can be viewed in all dimensions concurrently. This makes some operations re. language simpler (e.g. finding translations is just a slice range), but the idea is that it should also be relatively inexpensive to add more dimensions if needed (e.g. role).
Fixes #10169
Fixes #10364
Fixes #10482
Fixes #10630
Fixes #10656
Fixes #10694
Fixes #10918
Fixes #11262
Fixes #11439
Fixes #11453
Fixes #11457
Fixes #11466
Fixes #11540
Fixes #11551
Fixes #11556
Fixes #11654
Fixes #11661
Fixes #11663
Fixes #11664
Fixes #11669
Fixes #11671
Fixes #11807
Fixes #11808
Fixes #11809
Fixes #11815
Fixes #11840
Fixes #11853
Fixes #11860
Fixes #11883
Fixes #11904
Fixes #7388
Fixes #7425
Fixes #7436
Fixes #7544
Fixes #7882
Fixes #7960
Fixes #8255
Fixes #8307
Fixes #8863
Fixes #8927
Fixes #9192
Fixes #9324
2023-12-24 13:11:05 -05:00
|
|
|
// The config to pass to Build.
|
2023-05-15 02:39:52 -04:00
|
|
|
BuildCfg BuildCfg
|
2022-02-09 07:41:04 -05:00
|
|
|
}
|