GH-132983: Use a standard autoconf version check for zstd - #133479
Conversation
Sorry, something went wrong.
|
🤖 New build scheduled with the buildbot fleet by @AA-Turner for commit afbc562 🤖 Results will be shown at: https://buildbot.python.org/all/#/grid?branch=refs%2Fpull%2F133479%2Fmerge If you want to schedule another build, you need to add the 🔨 test-with-buildbots label again. |
Sorry, something went wrong.
|
@AA-Turner That looks promising: it seems to handle my use case. Thanks! cc: @emmatyping |
Sorry, something went wrong.
|
... and the ghcr.io/cirruslabs/macos-runner:sonoma runner also now appears to successfully build _zstd. |
Sorry, something went wrong.
emmatyping
left a comment
There was a problem hiding this comment.
Thanks! I didn't know this was something one could do. TIL
I'd like to wait for a RHEL8 buildbot to go green before merging to ensure that it works when there is a version of libzstd that is too old. Otherwise it looks much better.
Sorry, something went wrong.
|
And thank you for catching the original issue @ned-deily ! |
Sorry, something went wrong.
There is always something in Autotools that no one knows you can do. |
Sorry, something went wrong.
|
https://buildbot.python.org/#/builders/170/builds/1765 checking for libzstd >= 1.4.5... no checking for zstd.h... no checking for zdict.h... no ... checking for stdlib extension module _zstd... missing Seems to work. A |
Sorry, something went wrong.
pythonGH-133479 removed the logic to check for libzstd outside of pkg-config. This commit adds that logic back with a check for the version so that users can provide their own libzstd. This is to bring parity with lzma, bz2, and zlib detection.
pythonGH-133479 removed the logic to check for libzstd outside of pkg-config. This commit adds that logic back with a check for the version so that users can provide their own libzstd. This is to bring parity with lzma, bz2, and zlib detection.
cc @ned-deily, it seems that this works & is simpler? I'll run buildbots.
A