Release doc page dates - #7849
Open
bmwiedemann wants to merge 2 commits into
Open
Conversation
generate_last_commit_file.py keyed core_modules_with_last_commit.json by the basename of each directory containing a .html file, while get_git_commit_from_file() looks entries up by page name. Pages whose name differs from their directory therefore never matched: r.watershed (html in front/), r3.mapcalc (in the r.mapcalc directory), the wxGUI.* pages (all in gui/wxpython/docs/), the intro and driver pages (databaseintro in db/, helptext in lib/init/, ...) and the Markdown-only development pages in doc/, which have no .html file at all. In a build from a release tarball these pages fell back to the source directory mtime, which is the build time, so they said "Accessed: <build time>" instead of "Latest change: <date> in commit: <hash>" and differed between two builds of the same source. Key the JSON by page name instead, one entry per .html or .md file, keeping the last commit of the directory the page lives in as the recorded commit so that existing entries and the output of builds with Git history are unchanged. Sort the walk so that the few duplicate page names (r.drain in raster/ and scripts/, the example pages under doc/) resolve to the same entry in every run. This grows the file from 586 to 651 entries. mkhtml.py had a second bug hiding some of the same pages: the "<!-- meta page name: -->" and "<!-- meta page description: -->" comments overwrote pgm with the page title (e.g. "LRS" for the lrs page), which is not a JSON key, so even pages with a matching entry missed the lookup. Keep the title in a separate pgm_title used for the generated header and source code section, whose rendered output is unchanged, and leave pgm as the page name for the commit and addon path lookups. Together with the release workflow, which already runs generate_last_commit_file.py and ships the JSON in the release tarball, this makes all the previously date-stamped pages (71 html, 71 man, 74 markdown in GRASS 8.5.0) show the real last-change date and build reproducibly. This was found and fixed while working on reproducible builds for openSUSE, with the analysis and patch drafted by an AI assistant (Claude) and reviewed by the author.
Images in manual_gallery.html were sorted by the manual page they belong to alone, so multiple images of the same page (e.g. the three wxGUI.tplot screenshots) kept their os.listdir() order, which is the filesystem order and differs between builds. Sort by image name within one manual page to make the gallery reproducible. This patch was done while working on reproducible builds for openSUSE.
echoix
reviewed
Aug 25, 2026
echoix
left a comment
echoix
left a comment
Member
There was a problem hiding this comment.
The end of the tests are doing something different, I still think it's a valid change. Can someone else review?
| all core modules with their last commit. Used by GitHub "Additional Checks" | ||
| action workflow. | ||
| Script for testing that the core_modules_with_last_commit.json file | ||
| contains every documentation page with its last commit. Used by GitHub |
Member
There was a problem hiding this comment.
To reread, it's different, but I like the new text better
Member
|
There seems to be an outage of some sort |
wenzeslaus
reviewed
Aug 26, 2026
wenzeslaus
left a comment
wenzeslaus
left a comment
Member
There was a problem hiding this comment.
This looks pretty good overall. I have a question about the test.
| core_module = os.path.basename(core_module_path) | ||
| assert core_module in read_json_file | ||
| @pytest.mark.parametrize(("page", "page_path"), PAGES) | ||
| def test_pages_in_json_file(read_json_file, page, page_path): |
Member
There was a problem hiding this comment.
Maybe I'm just confused about fixtures here, but page_path is unused.
Contributor
Author
There was a problem hiding this comment.
It is used in test_compare_json_file_data and this way, we can use the same list of PAGES for both.
Contributor
Author
There was a problem hiding this comment.
There is way to not have the unused param, without duplicating the list:
@pytest.mark.depends(on=["test_json_file_is_not_empty"])
@pytest.mark.parametrize("page", [page for page, _ in PAGES])
def test_pages_in_json_file(read_json_file, page):
assert page in read_json_filebut not sure, if it is worth the change.
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
3 participants
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
While working on reproducible builds for openSUSE, I found that
our
grass-8.5.0package still had variations ofAccesseddate in html and md docs.I had ClaudeAI debug it and it found that the replacements of #3417 were not enough:
A) mapping of doc entries to JSON was faulty and incomplete
B) ordering influenced results
C) the fallback of using directory mtimes causes variations, because extracting a tarball updates these dir-mtimes.
Some testing was done and looked good.