Remember how #PyYAML ignored #Cython 3 compatibility problems for a year and a half? Well, Cython 3 got released today and now the related bug exploded when tons of pipelines suddenly started failing.
Remember how #PyYAML ignored #Cython 3 compatibility problems for a year and a half? Well, Cython 3 got released today and now the related bug exploded when tons of pipelines suddenly started failing. Today's #WTF: how threadpoolctl (a #Python package) asserts that the tests are run on a specific CPU. Sigh. Of course I've noticed because my CPU is not on the list! I wonder if we should start filing bugs about missing CPUs until upstream realizes it's a bad idea. https://github.com/joblib/threadpoolctl/blob/master/tests/test_threadpoolctl.py#L599-L622 Here's an idea: hibernate through the whole allergy/heat period, and get back to living around fall. In other #Python news, the new #scikit-build-core #pep517 backend turns out to break building extensions via #setuptools in yet another way, this time noticed via #pdm-backend. Oh, and there's also this bug where it breaks fail2ban too. Perhaps it's the same issue. scikit-build-core: https://github.com/scikit-build/scikit-build-core/issues/426 Remember when I said a week ago that #LLVM 17.x is nearing the branch point and it's looking good so far? Well, #Clang was broken 5 commits later, they knew it and disregarded is as "happens on just one platform". Well, surprise, it's not "just one platform", and the code is definitely broken somewhere! https://reviews.llvm.org/D143241#4476145 Nothing better to ruin your (already bad) mood in the morning than some obstinate #Python developer that insists that others should fix and maintain a backport package forever because they can't have literally a single `if` condition in their precious package. @mgorny TBH the discussion doesn't seem bad to me. They don't say "fuck off", just ask a few questions. I've seen discussions way worse, especially in PyCQA and PyPA projects where dramas are a daily routine. FOSS maintainers live in a different world. Our "current" #python version is 3.8. The 3.10 version reaches EoL in Ocy 2026, and that's when we'll be able to unconditionally use ExceptionGroup. So, yeah, we're careful about using anything that isn't in 3.8 🙃 "Their #RSA key is bigger! I want a huge key as well!" https://lists.gnupg.org/pipermail/gnupg-users/2023-July/066627.html @mgorny Yep, saw that… and immediately thought: "this isn't an e-penis size competition." Only thing I use RSA for is legacy system support. My first work YubiKey I set up as RSA4096 because I wasn't sure all our systems support ECC. Personal YubiKey was similarly set up as RSA4096 for backward-compatibility. After breaking work YubiKey #1, I buy a new one, and set it up Curve25519. So far, if its ≥ 2014, it just works. Only Ubuntu 10.04 and 12.04 is a problem. @mgorny washing dishes under running water requires copious amounts of detergent aswell. @mgorny pre-rinse? My parents had a dishwasher in like 2008+ and it didn't need that and stuff being badly cleaned was pretty rare.
But well closed sink here so far, annoying part is more that kitchens are terribly badly designed, the bottom of the sink should be leveled with the counter. A random relationship thought: For two programs to be meaningfully called "compatible" with each other, it takes more than for them to be able to run simultaneously on the same computer. That said, these days even that is becoming hard.
Show previous comments
@mgorny you forgot "we discontinued this feature and you can't install old versions but don't worry we also sell this other thing with those features in" 🥲 We seem to be having #CryptoBros in #Gentoo, and they're not happy about my woke "communist propaganda".
Show previous comments
@mgorny the way Rust integration was added is fantastic. Best way of handling it among Linux distros. And that's coming from someone who writes and deploys tons of Rust code. |
The #PyYAML problem is only getting better. People are trying really hard to make their things work again. Apparently there's a way to force constraints on pip but it's not a perfect solution either.
I have two thoughts about this:
1. *Finally*, it's not just us in #Gentoo being hit by constraint conflicts (i.e. two packages in depgraph requiring different #Cython versions).
2. Given that PyYAML knew about the problem for a year and a half and did nothing to avoid it… do you think it's a good idea to continue using that package? Or #YAML in particular, Norway?
https://discuss.python.org/t/no-way-to-pin-build-dependencies/29833
#Python
The #PyYAML problem is only getting better. People are trying really hard to make their things work again. Apparently there's a way to force constraints on pip but it's not a perfect solution either.
I have two thoughts about this:
1. *Finally*, it's not just us in #Gentoo being hit by constraint conflicts (i.e. two packages in depgraph requiring different #Cython versions).