incredible find just now from when i was young and innocent and TLS errors meant nothing to me https://github.com/pantsbuild/pants/pull/7619

context: centos6 is the docker image that defines the manylinux2010 standard which is the only way you can upload linux wheels to pypi. pypi of course can only reject PEPs, never accept them.

@hipsterelectron isn't manylinux2010 eol in favor of manylinux_2_28 based on rhel 8?
@leo this happened because i posted at least 20 separate times on my pip PRs that this fucking guy swooped in to stonewall that "hey isn't it weird how our linux ABI is defined in terms of an EOL centos image" https://peps.python.org/pep-0599/ and his eyes suddenly flashed with recognition
PEP 599 – The manylinux2014 Platform Tag | peps.python.org

This PEP proposes the creation of a manylinux2014 platform tag to succeed the manylinux2010 tag introduced by PEP 513. It also proposes that PyPI and pip both be updated to support uploading, downloading, and installing manylinux2014 distributions on co...

Python Enhancement Proposals (PEPs)

@leo why is is called manylinux2014 when it was proposed in 2019? it is a mystery

Package authors should not upload non-compliant manylinux2014 wheels to PyPI, and should be aware that PyPI may begin blocking non-compliant wheels from being uploaded.

we created this complete fiction of a problem and you need to use this exact root fs to be allowed into our gated community

@leo one of the two people very much in charge of how linux is allowed to interface with the python community https://mail.python.org/archives/list/[email protected]/message/TKJA5VHQLNYOF5RD6EZVI4TBPRJWQRQU/

or a whole bunch of other super
arcane traps that we've discovered over time (e.g. the Python used for
the build has to be linked using the correct configure options)

Mailman 3 [Distutils] Re: Idea: perennial manylinux tag - Distutils-SIG - python.org

@leo this guy is also in charge of explaining why pypi can't do anything useful

All the complicated bit is in the docker image/auditwheel. But, for
these, it turns out the spec doesn't actually matter that much.

@hipsterelectron "This will allow package maintainers to take advantage of new tags more quickly, while making better use of limited volunteer time." lol

@leo consider the first manylinux PEP https://peps.python.org/pep-0513/#compilation-of-compliant-wheels
in 2016:

For Linux, the situation is much more delicate. In general, compiled Python extension modules built on one Linux distribution will not work on other Linux distributions, or even on different machines running the same Linux distribution with different system libraries installed.

https://peps.python.org/pep-0384/
now consider 2009:

On Linux, changes to the ABI are often not much of a problem: the system will provide a default Python installation, and many extension modules are already provided pre-compiled for that version.

PEP 513 – A Platform Tag for Portable Linux Built Distributions | peps.python.org

This PEP proposes the creation of a new platform tag for Python package built distributions, such as wheels, called manylinux1_{x86_64,i686} with external dependencies limited to a standardized, restricted subset of the Linux kernel and core userspace A...

Python Enhancement Proposals (PEPs)
@leo btw i never heard that pypi legit banned noncompliant wheels. i thought it was just a standard you did because fucking whatever. i was convinced for many years that ABI was a spooky scary skeleton. i got over that when it turns out tensorflow just lies all the time https://github.com/pantsbuild/pants/pull/7424 lies to pypi, lies to you, lies to itself
hack default compiler option sets to fix tensorflow 1.13.1 on python 3.7 by cosmicexplorer · Pull Request #7424 · pantsbuild/pants

See tensorflow/tensorflow#27078 for an upstream tensorflow issue I made asking about the cause of #7417. Problem Resolves #7417. tensorflow==1.13.1 uses the newer C++ ABI for the shared library in ...

GitHub
@leo github now serving me 500s instead of timing me out. they can't possibly understand how much this inflates my ego
@hipsterelectron no no this is normal github's uptime has been hovering around and occasionally dipping under 90% the past few months
@leo when will github and gnu savannah achieve an equiliibrium point

@leo i read an fsf letter a few days ago that said:

  • anubis is malware.......because it is javascript
  • we have two people handling everything. why are you looking at me like that. anyway we have two people
  • PLEASE DONATE

democratic party

@leo worst part of it was calling javascript malware and not the browser. i refuse to use ableist insults but i have decided gnu is this little pocket universe and the fsf is just some letters
@hipsterelectron @leo LMAO are they STILL trying to pull that bullsnit??? uwu JS is so Bad and Evil because I Didn't Choose To Run That Code... even though I initiated the browser request to the site that served it to me 👉👈 uwu
@astraleureka @leo nothing has ever made me feel more immediately nihilistic
@astraleureka @leo gnu makes a lot of good software but i want to make a successor specifically to the fsf

@leo oh wait yeah here's the good shit from that first manylinux PEP

Build tools using PEP 425 platform tags do not track information about the particular Linux distribution or installed system libraries, and instead assign all wheels the too-vague linux_i686 or linux_x86_64 tags.

we have defined a problem!

It would be ideal if wheel packages could be compiled that would work on any linux system.

oh???? oh????

But,

boom. they never see it coming

Instead, we define a standard subset of the kernel+core userspace ABI

this part is the prestige. hit em with the smoke and sonics, then follow up with the generics

@leo no ABI was harmed in the making of this wheel.