Test GLIBC compatibility & Use GLIBC 2.28 w/Linux ARM #34

conda-forge-admin · 2025-01-27T19:31:53Z

Make top-level package name distinct (some reorg follows)
Use GLIBC 2.28 for Linux ARM builds
Use {{ stdlib("c") }} in the compilation test (and drop a related workaround)
Run check-glibc
Drop os_version (as this is sufficiently handled by the above changes)

Fixes #13

conda-forge-admin · 2025-01-27T19:33:16Z

Hi! This is the friendly automated conda-forge-linting service.

I just wanted to let you know that I linted all conda-recipes in your PR (recipe/meta.yaml) and found it was in an excellent condition.

I do have some suggestions for making it better though...

For recipe/meta.yaml:

ℹ️ The recipe is not parsable by parser conda-souschef (grayskull). This parser is not currently used by conda-forge, but may be in the future. We are collecting information to see which recipes are compatible with grayskull.

_{This message was generated by GitHub Actions workflow run https://github.com/conda-forge/conda-forge-webservices/actions/runs/13003589932. Examine the logs at this URL for more detail.}

…nda-forge-pinning 2025.01.27.18.10.49

jakirkham · 2025-01-27T20:18:56Z

@conda-forge-admin , please re-render

…nda-forge-pinning 2025.01.27.18.10.49

Rebuild now that Linux ARM uses GLIBC 2.28.

jakirkham · 2025-01-27T20:24:22Z

@carterbox could you please review?

carterbox · 2025-01-27T22:44:14Z

We're getting unexpected behavior because the top-level and first output are merged. i.e.

/home/conda/feedstock_root/build_artifacts/libcublas_1738017321092/work/conda_build.sh: line 7: check-glibc: command not found

carterbox · 2025-01-27T22:46:14Z

Demerging those outputs causes(?) the tests for the -dev package to fail.

Since the top-level package name now is `libcublas-split`, we cannot use `PKG_NAME` in the build script. So add a new variant `component_name` to `conda_build_config.yaml` and use that throughout.

jakirkham · 2025-01-28T03:17:17Z

@conda-forge-admin , please re-render

jakirkham · 2025-01-28T03:19:29Z

recipe/build.sh

    else
        # Put all other files in targetsDir
-        mkdir -p ${PREFIX}/${targetsDir}/${PKG_NAME}
-        cp -rv $i ${PREFIX}/${targetsDir}/${PKG_NAME}
+        mkdir -p ${PREFIX}/${targetsDir}/${component_name}
+        cp -rv $i ${PREFIX}/${targetsDir}/${component_name}


We were relying on the PKG_NAME to add some extra bits here. However now they go in libcublas-split (given that is when this build script runs

Have tried to address this here by creating a new variant/variable to handle this install location

…nda-forge-pinning 2025.01.27.22.18.09

jakirkham · 2025-01-28T03:26:37Z

@conda-forge-admin , please re-render

…nda-forge-pinning 2025.01.27.22.18.09

Stick to a unique variable name instead of clobbering the conda-build value. This is clearer long term.

jakirkham · 2025-01-28T03:39:15Z

@conda-forge-admin , please re-render

conda-forge-admin · 2025-01-28T03:41:28Z

Hi! This is the friendly automated conda-forge-webservice.

I tried to rerender for you, but it looks like there was nothing to do.

_{This message was generated by GitHub Actions workflow run https://github.com/conda-forge/conda-forge-webservices/actions/runs/13003174999. Examine the logs at this URL for more detail.}

jakirkham · 2025-01-28T03:45:43Z

recipe/build.sh

+# Run GLIBC check
+check-glibc bin/* lib/*.so.* lib64/*.so.*


Is this ok for a check? Or do we need to do anything additional here?

The `RECIPE_ROOT` variable was not set after moving the tests into the package name. However this is not needed anyways as we already instruct `conda-build` to copy the test file into the test directory. So it should already be in the current working directory. Given this, just drop it.

This fixes an issue where the GLIBC 2.17 sysroot would otherwise be pulled in.

jakirkham · 2025-01-28T04:11:06Z

@conda-forge-admin , please re-render

conda-forge-admin · 2025-01-28T04:13:16Z

Hi! This is the friendly automated conda-forge-webservice.

I tried to rerender for you, but it looks like there was nothing to do.

_{This message was generated by GitHub Actions workflow run https://github.com/conda-forge/conda-forge-webservices/actions/runs/13003456521. Examine the logs at this URL for more detail.}

carterbox

LGTM assuming that the missing DSO warnings about libgcc and stdlibc++ are expected due to the targets directory.

https://dev.azure.com/conda-forge/feedstock-builds/_build/results?buildId=1162787&view=logs&j=7b6f2c87-f3a7-5133-8d84-7c03a75d9dfc&t=9eb77fd2-8ddd-5444-8fc0-71cb28dcb736&l=987

jakirkham · 2025-01-28T17:48:46Z

Thanks Daniel! 🙏

Yep these are not new warnings. For example we see the same thing on main

dummy commit for rerendering

71b9d5f

conda-forge-webservices bot mentioned this pull request Jan 27, 2025

libcublas on aarch64 requires glibc 2.27 #13

Closed

MNT: Re-rendered with conda-build 25.1.1, conda-smithy 3.45.4, and co…

0ccda1a

…nda-forge-pinning 2025.01.27.18.10.49

conda-forge-admin marked this pull request as ready for review January 27, 2025 19:34

conda-forge-admin requested review from a team as code owners January 27, 2025 19:34

jakirkham added 2 commits January 27, 2025 12:11

Use GLIBC 2.28 on Linux ARM

1582e17

Drop sysroot GLIBC 2.28 workaround

aa4fb07

jakirkham changed the title ~~MNT: rerender~~ Use GLIBC 2.28 w/Linux ARM Jan 27, 2025

Check GLIBC compatibility

3869b12

jakirkham force-pushed the conda_forge_admin_13 branch from d8e397f to 3869b12 Compare January 27, 2025 20:18

conda-forge-webservices[bot] and others added 2 commits January 27, 2025 20:21

MNT: Re-rendered with conda-build 25.1.1, conda-smithy 3.45.4, and co…

64a566f

…nda-forge-pinning 2025.01.27.18.10.49

Bump build/number to 1

39bf5b5

Rebuild now that Linux ARM uses GLIBC 2.28.

jakirkham mentioned this pull request Jan 27, 2025

Update for CUDA 12.8.0 #33

Merged

carterbox added 2 commits January 27, 2025 15:21

BLD: Can't test a package which requires glibc 2.28 on cos7

ae88637

REF: Demerge top-level and first output

d64b7b1

carterbox force-pushed the conda_forge_admin_13 branch from d64b7b1 to ae88637 Compare January 27, 2025 22:33

jakirkham added 2 commits January 27, 2025 19:05

Add compilers & others to top-level requirements

2274584

Use component_name for feedstock & pkg extras

4479f24

Since the top-level package name now is `libcublas-split`, we cannot use `PKG_NAME` in the build script. So add a new variant `component_name` to `conda_build_config.yaml` and use that throughout.

jakirkham reviewed Jan 28, 2025

View reviewed changes

conda-forge-webservices[bot] and others added 2 commits January 28, 2025 03:19

MNT: Re-rendered with conda-build 25.1.1, conda-smithy 3.45.4, and co…

0ee194c

…nda-forge-pinning 2025.01.27.22.18.09

Override PKG_NAME with name

a805ff2

MNT: Re-rendered with conda-build 25.1.1, conda-smithy 3.45.4, and co…

e37d01e

…nda-forge-pinning 2025.01.27.22.18.09

Replace PKG_NAME w/component_name again

91c4e81

Stick to a unique variable name instead of clobbering the conda-build value. This is clearer long term.

jakirkham reviewed Jan 28, 2025

View reviewed changes

jakirkham added 3 commits January 27, 2025 19:58

Drop compilers from top-level requirements

98b949d

Add {{ stdlib("c") }} in compiler test

d84f2f9

This fixes an issue where the GLIBC 2.17 sysroot would otherwise be pulled in.

jakirkham added 2 commits January 27, 2025 20:24

Match bool casing in recipe

1684934

Disable binary relocation in libcublas-static

2063f28

jakirkham requested a review from carterbox January 28, 2025 04:26

jakirkham mentioned this pull request Jan 28, 2025

Configure and Test GLIBC versions conda-forge/libcusolver-feedstock#23

Closed

jakirkham changed the title ~~Use GLIBC 2.28 w/Linux ARM~~ Test GLIBC compatibility & Use GLIBC 2.28 w/Linux ARM Jan 28, 2025

jakirkham mentioned this pull request Jan 28, 2025

Test GLIBC compatibility & Use GLIBC 2.28 w/Linux ARM conda-forge/libcusolver-feedstock#24

Merged

carterbox approved these changes Jan 28, 2025

View reviewed changes

jakirkham merged commit be9bb52 into conda-forge:main Jan 28, 2025
6 checks passed

jakirkham mentioned this pull request Jan 28, 2025

cusparselt v0.7.0.0 conda-forge/cusparselt-feedstock#52

Merged

3 tasks

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Test GLIBC compatibility & Use GLIBC 2.28 w/Linux ARM #34

Test GLIBC compatibility & Use GLIBC 2.28 w/Linux ARM #34

conda-forge-admin commented Jan 27, 2025 •

edited by jakirkham

Loading

conda-forge-admin commented Jan 27, 2025 •

edited

Loading

jakirkham commented Jan 27, 2025

jakirkham commented Jan 27, 2025

carterbox commented Jan 27, 2025

carterbox commented Jan 27, 2025

jakirkham commented Jan 28, 2025

jakirkham Jan 28, 2025

jakirkham commented Jan 28, 2025

jakirkham commented Jan 28, 2025

conda-forge-admin commented Jan 28, 2025

jakirkham Jan 28, 2025

carterbox Jan 28, 2025

jakirkham commented Jan 28, 2025

conda-forge-admin commented Jan 28, 2025

carterbox left a comment

jakirkham commented Jan 28, 2025

Test GLIBC compatibility & Use GLIBC 2.28 w/Linux ARM #34

Test GLIBC compatibility & Use GLIBC 2.28 w/Linux ARM #34

Conversation

conda-forge-admin commented Jan 27, 2025 • edited by jakirkham Loading

conda-forge-admin commented Jan 27, 2025 • edited Loading

jakirkham commented Jan 27, 2025

jakirkham commented Jan 27, 2025

carterbox commented Jan 27, 2025

carterbox commented Jan 27, 2025

jakirkham commented Jan 28, 2025

jakirkham Jan 28, 2025

Choose a reason for hiding this comment

jakirkham commented Jan 28, 2025

jakirkham commented Jan 28, 2025

conda-forge-admin commented Jan 28, 2025

jakirkham Jan 28, 2025

Choose a reason for hiding this comment

carterbox Jan 28, 2025

Choose a reason for hiding this comment

jakirkham commented Jan 28, 2025

conda-forge-admin commented Jan 28, 2025

carterbox left a comment

Choose a reason for hiding this comment

jakirkham commented Jan 28, 2025

conda-forge-admin commented Jan 27, 2025 •

edited by jakirkham

Loading

conda-forge-admin commented Jan 27, 2025 •

edited

Loading