Created attachment 257571 [details] textproc/py-sphinx: Update to 8.1.3 poudriere-testport textproc/py-sphinx: main-amd64-current: Pass main-i386-current: Pass 13_4-amd64-release: Pass 13_4-i386-release: Pass 14_1-amd64-release: Pass 14_1-i386-release: Pass 14_2-amd64-release: Pass 14_2-i386-release: Pass
Hello,
Comment on attachment 257571 [details] textproc/py-sphinx: Update to 8.1.3 Update patch for 8.2.0 is almost complete with poudriere-testport, will upload then.
Created attachment 257653 [details] textproc/py-sphinx: Update to 8.2.0 Update to 8.1.3: * Remove ${DISTNAME} * Remove ${EXTRACT_CMD} as upstream issue has been resolved * Remove Python <3.10 check * Remove files/patch-pyproject.toml * Some port{clippy,fmt,lint} maintenance * Switch to <bsd.port.mk> inclusion declaration * Use ${DISTVERSION} over ${PORTVERSION} * Use ${PEP517_BUILD_*} Helpers Unable to run make test, because devel/py-setuptools (bug #270358) is outdate. poudriere-testport textproc/py-sphinx: main-amd64-default: Pass main-i386-default: Pass 13_4-amd64-release: Pass 13_4-i386-release: Pass 14_1-amd64-release: Pass 14_1-i386-release: Pass 14_2-amd64-release: Pass 14_2-i386-release: Pass
Created attachment 257668 [details] textproc/py-sphinx: Update to 8.2.0 s/8.1.3/8.2.0/ in commit log
@rene please clear the exp-run if it does not need an exp-run , otherwise reassign to portmgr
Ah, my bad, I was confused.
Comment on attachment 257668 [details] textproc/py-sphinx: Update to 8.2.0 Will update with a new patch for 8.2.1
Created attachment 257759 [details] textproc/py-sphinx: Update to 8.2.1 * Remove ${DISTNAME} * Remove ${EXTRACT_CMD} as upstream issue has been resolved * Remove Python <3.10 check * Remove files/patch-pyproject.toml * Some port{clippy,fmt,lint} maintenance * Switch to <bsd.port.mk> inclusion declaration * Use ${DISTVERSION} over ${PORTVERSION} * Use ${PEP517_BUILD_*} Helpers poudriere-testport textproc/py-sphinx: main-amd64-default: Pass main-i386-default: Pass 13_4-amd64-release: Pass 13_4-i386-release: Pass 14_1-amd64-release: Pass 14_1-i386-release: Pass 14_2-amd64-release: Pass 14_2-i386-release: Pass make test (15-CURRENT + local py-setuptool-75 patch): 4 failed, 2293 passed, 31 skipped, 5 deselected, 2 xfailed
Some new failure logs: https://pkg-status.freebsd.org/gohan04/data/141amd64-default-foo/2025-02-22_19h10m05s/logs/errors/cmake-doc-3.31.5.log https://pkg-status.freebsd.org/gohan04/data/141amd64-default-foo/2025-02-22_19h10m05s/logs/errors/colmap-3.11.1_1.log https://pkg-status.freebsd.org/gohan04/data/141amd64-default-foo/2025-02-22_19h10m05s/logs/errors/ghc-9.6.6_2.log https://pkg-status.freebsd.org/gohan04/data/141amd64-default-foo/2025-02-22_19h10m05s/logs/errors/gnuradio-3.8.4.0_24.log https://pkg-status.freebsd.org/gohan04/data/141amd64-default-foo/2025-02-22_19h10m05s/logs/errors/kitty-0.39.1.log https://pkg-status.freebsd.org/gohan04/data/141amd64-default-foo/2025-02-22_19h10m05s/logs/errors/libbson-1.25.4.log https://pkg-status.freebsd.org/gohan04/data/141amd64-default-foo/2025-02-22_19h10m05s/logs/errors/pocl-4.0.log https://pkg-status.freebsd.org/gohan04/data/141amd64-default-foo/2025-02-22_19h10m05s/logs/errors/py311-boolean.py-4.0.log https://pkg-status.freebsd.org/gohan04/data/141amd64-default-foo/2025-02-22_19h10m05s/logs/errors/py311-breathe-4.35.0_2.log https://pkg-status.freebsd.org/gohan04/data/141amd64-default-foo/2025-02-22_19h10m05s/logs/errors/py311-django-configurations-2.5_1.log https://pkg-status.freebsd.org/gohan04/data/141amd64-default-foo/2025-02-22_19h10m05s/logs/errors/py311-django-easy-thumbnails-2.8.1_2.log https://pkg-status.freebsd.org/gohan04/data/141amd64-default-foo/2025-02-22_19h10m05s/logs/errors/py311-django-reversion-5.0.10_1.log https://pkg-status.freebsd.org/gohan04/data/141amd64-default-foo/2025-02-22_19h10m05s/logs/errors/py311-django-simple-captcha-0.5.20_1.log https://pkg-status.freebsd.org/gohan04/data/141amd64-default-foo/2025-02-22_19h10m05s/logs/errors/py311-django-treebeard-4.7.1_1.log https://pkg-status.freebsd.org/gohan04/data/141amd64-default-foo/2025-02-22_19h10m05s/logs/errors/py311-gntp-1.0.3.log https://pkg-status.freebsd.org/gohan04/data/141amd64-default-foo/2025-02-22_19h10m05s/logs/errors/py311-khard-0.19.1.log https://pkg-status.freebsd.org/gohan04/data/141amd64-default-foo/2025-02-22_19h10m05s/logs/errors/py311-kivy-2.3.0.log https://pkg-status.freebsd.org/gohan04/data/141amd64-default-foo/2025-02-22_19h10m05s/logs/errors/py311-notmuch-0.38.3.log https://pkg-status.freebsd.org/gohan04/data/141amd64-default-foo/2025-02-22_19h10m05s/logs/errors/py311-offlineimap3-8.0.0.70.log https://pkg-status.freebsd.org/gohan04/data/141amd64-default-foo/2025-02-22_19h10m05s/logs/errors/py311-paperless-2.7.0_19.log https://pkg-status.freebsd.org/gohan04/data/141amd64-default-foo/2025-02-22_19h10m05s/logs/errors/py311-pip-23.3.2_3.log https://pkg-status.freebsd.org/gohan04/data/141amd64-default-foo/2025-02-22_19h10m05s/logs/errors/py311-pynetbox-7.4.1.log https://pkg-status.freebsd.org/gohan04/data/141amd64-default-foo/2025-02-22_19h10m05s/logs/errors/py311-sorl-thumbnail-12.10.0_1.log https://pkg-status.freebsd.org/gohan04/data/141amd64-default-foo/2025-02-22_19h10m05s/logs/errors/py311-virtualenvwrapper-4.8.4_1.log https://pkg-status.freebsd.org/gohan04/data/141amd64-default-foo/2025-02-22_19h10m05s/logs/errors/py311-webob-1.8.7.log https://pkg-status.freebsd.org/gohan04/data/141amd64-default-foo/2025-02-22_19h10m05s/logs/errors/srain-1.7.0.log https://pkg-status.freebsd.org/gohan04/data/141amd64-default-foo/2025-02-22_19h10m05s/logs/errors/xca-2.8.0.log https://pkg-status.freebsd.org/gohan04/data/141amd64-default-foo/2025-02-22_19h10m05s/logs/errors/xca-qt6-2.8.0.log py311-kivy kills the disk space of the builders
(In reply to Antoine Brodin from comment #10) Thanks. I Will start working on some these in a couple of days.
I have not forgotten about the list of work here. I have two Ports working in progress, pip, and numpy, however, I am not happy with my work, and the test requirements, require loads more Ports.
^Triage: reassign back to ports-bugs@ now that portmgr has completed the exp-run and has asked for feedback.
Can we get this moving again? The current sphinx port (it's epub3 module) imports imghdr, which was removed in Python 3.13, so this is getting kinda urgent so we can move the many ports using sphinx somehow to Python 3.13 or 3.14. Python 3.9 was removed and laid to rest a few hours ago.
(In reply to Matthias Andree from comment #14) If it's too difficult to do the whole exp-run and fix all the broken dependent ports, we could also attempt to import this sphinx fix to work around the imghdr removal: https://github.com/sphinx-doc/sphinx/commit/a502e7523376e0344c1c9cc8a9d128143cd98b2d
What isn't working with the patches on this ticket, and where is help required? Does anything more break on the current version of Sphinx (9.0.4)?
Created attachment 267536 [details] py-sphinx-9.1.0,1.patch Update to 9.1.0 https://www.sphinx-doc.org/en/master/changes/#release-9-1-0-released-dec-31-2025
It seems that if the cleanup up to python-3.11 cannot be completed, instead of updating, some work resembling PY_SETUPTOOLS will likely be needed in Mk/Uses/python.mk...
Created attachment 269524 [details] py314-sphinx-9.1.0,1.patch (In reply to Tatsuki Makino from comment #18) Many people are probably having trouble updating their ports to the latest version because of this issue :-( As a side note, I've corrected an error in the patch for 9.1.0 that I sent out recently. The updates are as follows: RUN_DEPENDS: Stricter version check TEST_DEPENDS: Streamlined EXTRACT_CMD: Executable type is now a variable portclippy(1): Compliant https://github.com/sphinx-doc/sphinx/releases/tag/v9.1.0 I hope it gets resolved soon :-)
I am trying to port pgadmin4, and stuck in need of newer sphinx. I am eager to do the work (may be not so frequetly responsive). I noticed some points maybe in need of discussion. * Recent patch to 9.1.0 requires python version 3.12 and above. While the default version of python in ports is still 3.11, should we stick to the version supporting python 3.11 as well? (8.x?) * The error logs from the exp-run (in 2025) seem to have dissapeared. Is it possible to redo the exp-run if I will work on it?
Created attachment 269546 [details] patch to py-sphix 8.1.3 As we have still python 3.10 in the ports, textproc/py-sphinx should support it. The latest version that supports python 3.10 was sphinx 8.1.3. What step should I take as next step?
(In reply to Hiroo Ono from comment #21) textproc/py-sphinx is a build tool. In general ports should not depend on it at runtime (though some ports do depend, they should be fixed), so in principle we could have multiple version of textproc/py-sphinx in the ports tree. But having multiple version for numerous textproc/py-sphinx-* modules would be overkill. I suggest to update textproc/py-sphinx to the last version which still supports Python 3.11, and remove textproc/py-sphinx from any consumer that does not support this version.
(In reply to Max Brazhnikov from comment #22) The latest version supporting python 3.11 is sphinx 9.0.4. I will remake the patch. However, if there are people who set python default version to 3.10 in their make.conf (I do not know how many there are), they will have trouble building the softwares containing sphinx documents. Should not we support them? Or, suggest them to set the DOCS option off for the ports with sphinx document and urge them to switch to higher version of python?
(In reply to Hiroo Ono from comment #23) Well, Python 3.10 is heading to EOL. If you want to support Python 3.10 nobody will stop you :)
Created attachment 269550 [details] patch to update to textproc/py-sphinx 9.0.4 Following the comment from Max Brazhnikov, I recreated the patch for sphinx 9.0.4.
Created attachment 269645 [details] Sphinx 9.0.4 patch (updated against recent ports)
Forks, I have created a branch for this update: https://codeberg.org/makc/freebsd-ports/src/branch/sphinx The current plan is to check and fix fallout from the previous exp-run and then I will request new exp-run. Help is appreciated!
About the subject, 9.1.0 does not support python3.11, is it OK? I am now carrying the poudriere bulk of the ports that have dependency on texproc/py-sphinx. Currently, emulators/mame is taking forever, but I think I can show you the logs in a day or two.
(In reply to Hiroo Ono from comment #28) poudriere bulk against sphinx 9.0.4 I meant.
For now, I found the ports below not to build against sphinx 9.0,4 and made update patches: bug 294473: devel/py-pip It claims to only support sphinx 7.*, but could be built against sphinx 9.0.4. Incompatible with current version of sphinx in ports (5.3.8) bug 294475: devel/py-dbus The dependent port textproc/py-sphinx_rtd_theme needs to be updated. bug 294474: textproc/py-sphinx_rtd_theme It needs sphinx >= 6, incompatible with current version of sphinx in ports (5.3.8)
(In reply to Hiroo Ono from comment #30) Please, avoid unneeded changes (including style changes) for these updates and fixes. If you want to preserve your name in the Author entry generate the patches with `git format-patch`. You may also add "PR: 284845" to the commit message. Attach all patches to this PR, no need to open separate PRs per patch. Thank you for your help!
We can't update to 9.1.0 due to dropped support for Python 3.11
(In reply to Max Brazhnikov from comment #31) Thank you for your comment. I will remake the patch. Is it better to send pull-request to your codeberg repository, or putting the patch in this PR is needed anyway?
https://barleycorn.oikumene.net/ports-patch/sphinx/2026-04-11_23h19m36s/build.html poudriere run of * textproc/py-sphinx updated to 9.0.4 * ports that have dependency on textproc/py-sphinx are built * SET= DOCS HTMLDOCS SPHINX is set in the make.conf I accdentally stopped the build while emulators/mame were still running. It is on 15-STABLE.
Please, instead of just changing the subject simultaneously, discuss which version of python we should support (support python 3.11 or not). I have no idea which to take, but I want that the version of sphinx I should check with be fixed.
Created attachment 269751 [details] [PATCH] textproc/py-sphinx: Update to 9.1.0 Update to 9.1.0: * Remove ${DISTNAME} * Remove ${EXTRACT_CMD} as upstream issue has been resolved * Remove files/patch-pyproject.toml * Switch to <bsd.port.mk> inclusion declaration * Use ${DISTVERSION} over ${PORTVERSION} * port{clippy,fmt,lint} maintenance poudriere-testport textproc/py-sphinx: 15_0-amd64-release: Pass main-amd64-default: Pass
(In reply to Matthias Andree from comment #14) I am testing (going thru the list from the previous exp-run) against Python-3.12, is something greater recommended?
(In reply to Dimitry Andric from comment #15) Covered in 9.1.0.
(In reply to Max Brazhnikov from comment #32) It looks like the switch to Python-3.12 as default should be ready, as numpy as has been updated. So why restrict another domain of the Ports Tree?
(In reply to Alastair Hogge from comment #39) I have no idea what is the status of the moving defaults to Python 3.12, so let's proceed step by step. However, feel free to open new PR for sphinx-9.1.0.
Antoine, I'd like to have exp-run for sphinx-9.0.4. The patch includes fixes for some ports from the previous exp-run, new textproc/py-sphinx5 for sphinx-5.3.0. Several ports from the previous exp-run are not fixed yet, but they should be updated to newer versions or stop generating docs via sphinx. I'll contact maintainers later. The patch is available at https://people.freebsd.org/~makc/patches/sphinx-9.0.4.diff or use https://codeberg.org/makc/freebsd-ports/src/branch/sphinx
(In reply to Hiroo Ono from comment #33) Either way is fine for me.
(In reply to Max Brazhnikov from comment #41) > https://people.freebsd.org/~makc/patches/sphinx-9.0.4.diff Bug #276417 (textproc/py-sphinx_rtd_theme), already modernises the Port, and correctly integrates the Port's documentation, why is a patch that meets a lesser standard being used? Regarding py-sphinx, my patch modernises the Port, and also removes the unnecessary ${EXTRACT_CMD} (https://github.com/sphinx-doc/sphinx/issues/2395), why does it remain in your patch? For decades, other volunteers' labour is continuously dismissed, ignored, or written off. Why do you continue to dissuade other from participating? I already turned down an offer to apply as a formal committer partially because of how you and others bully other volunteer/users. In fact bully is not the correct term, because bullies have a capacity for change, and all I observe is cowards avoiding dealing with contradictions, the puerile politics of conservatives crying feelings and vibes, and ego, over the greater aims of the project.
lang/python310 is still present in porttree, so it seems necessary to settle on a version that can still be used at least once. I continued to use Python 3.9 even after Python 3.11 became the default, but it was difficult because ports that no longer worked with Python 3.9 had to be kept with pkg-lock :) What kind of method would be best? Are variables like setuptools provided? Are option switches created like doxygen...?
(In reply to Alastair Hogge from comment #43) > Bug #276417 (textproc/py-sphinx_rtd_theme), already modernises the Port, and > correctly integrates the Port's documentation, why is a patch that meets a > lesser standard being used? Sorry, I missed your PR. I'll check it. > Regarding py-sphinx, my patch modernises the Port, and also removes the > unnecessary ${EXTRACT_CMD} (https://github.com/sphinx-doc/sphinx/issues/2395), > why does it remain in your patch? I used the other patch which updates sphinx to 9.0.4. I'll appreciate if you prepare the patch for 9.0.4 with all your improvements to the port. > For decades, other volunteers' labour is continuously dismissed, ignored, or > written off. Why do you continue to dissuade other from participating? I already > turned down an offer to apply as a formal committer partially because of how you > and others bully other volunteer/users. In fact bully is not the correct term, > because bullies have a capacity for change, and all I observe is cowards > avoiding dealing with contradictions, the puerile politics of conservatives > crying feelings and vibes, and ego, over the greater aims of the project. Alastair, I'm sorry for your experience, but I'm not from the python@ team. I have recently dived into python ports only because I'm interested in having modern numpy/scipy in the ports tree. Btw, all ports still use numpy1, so the migration to numpy2 is still ongoing, see bug 294328
(In reply to Tatsuki Makino from comment #44) If you use non-default Python you build packages yourself I suppose. In this case you can disable DOCS option for ports which rely on sphinx. I'm not sure about introducing special option for sphinx generated docs. Doxygen port depends on heavy TeX packages, that was the reason to have dedicated DOXYGEN option.
Created attachment 269830 [details] [PATCH] textproc/py-sphinx: Update to 9.0.4 Update to 9.0.4: * Remove ${DISTNAME} * Remove ${EXTRACT_CMD} as upstream issue has been resolved * Remove files/patch-pyproject.toml * Switch to <bsd.port.mk> inclusion declaration * Use ${DISTVERSION} over ${PORTVERSION} * port{clippy,fmt,lint} maintenance https://www.sphinx-doc.org/en/master/changes/9.0.html#release-9-0-4-released-dec-04-2025 poudriere-testport textproc/py-sphinx: 15_0-amd64-release: Pass main-amd64-default: Pass