The matplotib vs meson-python issue is pure madness, see https://github.com/sagemath/sage/pull/39789
The problem is not that one cannot build matplotlib spkg. It does work on systems we tested, and the release manager agreed. That something does not match on PyPI is not an argument. We don't use meson-python to build matplotlib. Yes, matplotlib we ship cannot be built with meson-python we ship, but it doesn't matter, as meson-python is not needed to build matplotlib. Having said this, the right way out is to vendor neither matpotlib, nor meson-python. Then you just get what PyPI offers, and it's not our headache any more. On Thu, Apr 24, 2025 at 9:20 AM Marc Culler <marc.cul...@gmail.com> wrote: > I have run into a similar issue with the matplotlib spkg. I don't think > the issue is with the download. The log file shows a few 404's but > eventually the download works and the hash is correct. I think the real > issue is that the matplotlib package requires a version of meson-python > which is older than the one which sage installs. I see the following in > the log file for the matrplotlib build: > > [spkg-install] < ERROR: Could not find a version that satisfies the > requirement meson- > [spkg-install] python<0.17.0,>=0.13.1 (from versions: none) > [spkg-install] < ERROR: No matching distribution found for > meson-python<0.17.0,>=0.13.1 > > Sage is installing meson-python 0.17.0 and that is too new for matplotlib > 3.10.1, which is the latest version of matplotlib on pypi. > > - Marc > > On Thursday, April 24, 2025 at 1:26:06 AM UTC-5 aram.derme...@gmail.com > wrote: > >> Yes, that's what I'm currently trying to do, but finding all the packages >> with wrong checksums is turning out to not be so fun. Have to keep >> rerunning `make` to try and see which package breaks next. >> >> 2 questions though: >> 1. Is there not currently a way to do a "check integrity" run or >> something? Where you first check the checksums of all files before trying >> to run make? >> 2. I think there's still a bug in the following from the log >> ``` >> ERROR [transfer|run:135]: [Errno socket error] [Errno 404] Not Found: '// >> github.com/sagemath/sage/releases/download/10.7/matplotlib-3.10.1.tar.gz' >> ``` >> Notice how when it's trying to pull, it's trying to pull from 10.7 and >> *not* from 10.7.beta1. I looked in `/src/bin/sage-version.sh` (which is >> auto-generated) and the SAGE_VERSION is correctly set to '10.7.beta1'. >> (/src/sage/VERSION.txt is also showing the correct version) So somewhere >> along the line, when it's trying to download the correct tarbell, it's not >> doing it properly. >> Is this intended? Or should I create an issue ticket for this? >> >> -Aram >> >> On Thu, 24 Apr 2025 at 08:14, enriqu...@gmail.com <enriqu...@gmail.com> >> wrote: >> >>> If I am not wrong this is caused because sometimes downloading files >>> fail. In my experience a direct download of the files, copied in the folder >>> upstream, and restarting make is enough. >>> >>> El miércoles, 23 de abril de 2025 a las 19:17:15 UTC+2, >>> aram.derme...@gmail.com escribió: >>> >>>> This definitely solves the sci-py problem! Thanks. >>>> >>>> I'm still getting hung up on the 2nd issue though where for some reason >>>> matplotlib just won't install. The terminal goes to something like: >>>> ``` >>>> [sagenb_export-3.3] [spkg-pipinst] Installing collected packages: >>>> sagenb_export >>>> [sagenb_export-3.3] [spkg-pipinst] changing mode of >>>> /sage/sagedev/local/var/lib/sage/venv-python3.12.5/bin/sagenb-export to 755 >>>> [sagenb_export-3.3] [spkg-pipinst] Successfully installed >>>> sagenb_export-3.3 >>>> [sagenb_export-3.3] Deleting build directory >>>> /sage/sagedev/local/var/lib/sage/venv-python3.12.5/var/tmp/sage/build/sagenb_export-3.3 >>>> [sagenb_export-3.3] Finished installing sagenb_export-3.3 >>>> ``` >>>> (each time something slightly different) and then it just stalls there. >>>> (I Left it like that for 4 hours and no change. Then when I ctrl+c to stop >>>> the running process matplotlib throws an error. >>>> >>>> -Aram >>>> >>>> On Wed, 23 Apr 2025 at 13:25, Eric Gourgoulhon <egourg...@gmail.com> >>>> wrote: >>>> >>>>> Hi, >>>>> >>>>> This line in the scipy log: >>>>> [spkg-install] g++: fatal error: Killed signal terminated program >>>>> cc1plus >>>>> points towards a memory issue. Could you increase the RAM devoted to >>>>> WSL2 or decrease the number of processes if you are performing a parallel >>>>> build (e.g use make -j4 instead of make -j8) ? >>>>> >>>>> Eric. >>>>> >>>>> >>>>> Le mercredi 23 avril 2025 à 11:41:37 UTC+2, aram.derme...@gmail.com a >>>>> écrit : >>>>> >>>>>> Hey all, >>>>>> >>>>>> I'm trying to update to the latest devel version in order to code >>>>>> some things up, but I can't seem to get the latest devel version "making" >>>>>> properly. Not sure if I'm doing something wrong, so I'm attaching my >>>>>> logs. >>>>>> >>>>>> - First time, I ran it, I got a scipy error (logs attached) and it >>>>>> wouldn't build >>>>>> - I tried running `make` again (without doing a `make distclean`) and >>>>>> the 2nd time scipy installs correctly, but then the instlaler just hangs >>>>>> and doesn't finish. >>>>>> -> When I `ctrl+c` to quit, it shows `matplotlib-3.10.1` errors, so >>>>>> seems like it's getting stuck in `matplotlib-3.10.1`. (Which I'm >>>>>> attaching >>>>>> the log file for) >>>>>> -> Analyzying the log file, looks like it can't properly get >>>>>> `matplotlib-3.10.1`. Version 10.6 doesn't have it (only has >>>>>> `matplotlib-3.8` and the URL it's trying to download from for 10.7 is >>>>>> wrong >>>>>> (It should be: >>>>>> https://github.com/sagemath/sage/releases/download/10.7.beta1/matplotlib-3.10.1.tar.gz >>>>>> ) >>>>>> >>>>>> I tried installing a fresh copy of sagedev (not my forked version) >>>>>> and a similar thing happens, but this time it failed when trying to >>>>>> install >>>>>> numpy. >>>>>> >>>>>> Wanted to see what the best way to approach fixing my development >>>>>> version is. >>>>>> >>>>>> Also, I'm thinking I should create a ticket along the lines of "if >>>>>> unable to download xxx, then an error should be thrown instead of just >>>>>> hanging and waiting forever", but before doing that I want to make sure >>>>>> that it is really a problem with the dev version and not a problem on my >>>>>> end (as my internet is a little weird here, so it could potentially be my >>>>>> internet being weird) >>>>>> >>>>>> Kindly, >>>>>> Aram >>>>>> >>>>> -- >>>>> You received this message because you are subscribed to the Google >>>>> Groups "sage-devel" group. >>>>> To unsubscribe from this group and stop receiving emails from it, send >>>>> an email to sage-devel+...@googlegroups.com. >>>>> To view this discussion visit >>>>> https://groups.google.com/d/msgid/sage-devel/53b4ddd4-6e55-44fa-b6b0-87f9ee66d4f1n%40googlegroups.com >>>>> <https://groups.google.com/d/msgid/sage-devel/53b4ddd4-6e55-44fa-b6b0-87f9ee66d4f1n%40googlegroups.com?utm_medium=email&utm_source=footer> >>>>> . >>>>> >>>> -- >>> You received this message because you are subscribed to the Google >>> Groups "sage-devel" group. >>> To unsubscribe from this group and stop receiving emails from it, send >>> an email to sage-devel+...@googlegroups.com. >>> >> To view this discussion visit >>> https://groups.google.com/d/msgid/sage-devel/128735af-d3cc-4db9-bf33-213017bbce7an%40googlegroups.com >>> <https://groups.google.com/d/msgid/sage-devel/128735af-d3cc-4db9-bf33-213017bbce7an%40googlegroups.com?utm_medium=email&utm_source=footer> >>> . >>> >> -- > You received this message because you are subscribed to the Google Groups > "sage-devel" group. > To unsubscribe from this group and stop receiving emails from it, send an > email to sage-devel+unsubscr...@googlegroups.com. > To view this discussion visit > https://groups.google.com/d/msgid/sage-devel/4af793eb-4481-4ed0-b35f-34d403d492acn%40googlegroups.com > <https://groups.google.com/d/msgid/sage-devel/4af793eb-4481-4ed0-b35f-34d403d492acn%40googlegroups.com?utm_medium=email&utm_source=footer> > . > -- You received this message because you are subscribed to the Google Groups "sage-devel" group. To unsubscribe from this group and stop receiving emails from it, send an email to sage-devel+unsubscr...@googlegroups.com. To view this discussion visit https://groups.google.com/d/msgid/sage-devel/CAAWYfq2U%3DEDO2cmgV8BXF%3DFPzfsznwE3D7RkviGdyGQp7Ums6g%40mail.gmail.com.
