tadeja commented on PR #50903:
URL: https://github.com/apache/arrow/pull/50903#issuecomment-5369951362
Checked - 4 nightly release jobs are passing now
`verify-rc-source-{cpp/integration/python/ruby}-linux-conda-latest-amd64`,
resolving to correct GCC variant `hc6a0c74_2` which depends on
`conda-gcc-specs` and installs it as it should be:
` + conda-gcc-specs 15.3.0
h8846f6e_2 conda-forge`
` + gcc 15.3.0
hc6a0c74_2 conda-forge`
https://github.com/ursacomputing/crossbow/actions/runs/32371903836/job/96434179230#step:5:1088
However, `conda-cpp` and `conda-integration` images continue to reuse the
cached Docker layer created before the [upstream
fix](https://github.com/conda-forge/ctng-compilers-feedstock/pull/224) :
```
#16 [ 5/11] RUN mamba install -q -y --file
arrow/ci/conda_env_cpp.txt --file arrow/ci/conda_env_gandiva.txt
compilers ...
#16 CACHED
```
[Docker Test
conda-cpp](https://github.com/ursacomputing/crossbow/actions/runs/32432598296/job/96627190083#step:6:179)
[Docker Test
conda-python-hdfs](https://github.com/ursacomputing/crossbow/actions/runs/32432594768/job/96627180477#step:6:192)
[AMD64 Conda C++
AVX2](https://github.com/apache/arrow/actions/runs/32421387463/job/96593973481#step:8:161)
[AMD64 Ubuntu
Meson](https://github.com/apache/arrow/actions/runs/32431969985/job/96625303986#step:6:159)
[AMD64 Conda Integration
Test](https://github.com/apache/arrow/actions/runs/32481273811/job/96767960887#step:13:141)
so these still lack `conda-gcc-specs` (and GCC does not automatically search
$CONDA_PREFIX/include and $CONDA_PREFIX/lib).
@raulcd, @kou is there another way to invalidate stale Docker layer
containing the broken gxx/gcc conda-forge resolution?
Otherwise I would reduce this PR to changes in
[conda-cpp.dockerfile](https://github.com/apache/arrow/pull/50903/changes#diff-77e83f3866a7b845e723c3eb031ba36dcafc14d833ace16baecaae1a2056cf9e)
and
[conda-integration.dockerfile](https://github.com/apache/arrow/pull/50903/changes#diff-d2694803c775ca15607d375cf0f3d942895c52159c26e30a1ea1e771a2be3ebf)
Once the cached images get rebuilt, we can decide to keep explicit install
of `conda-cpp-specs` or rely on the corrected upstream.
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
To unsubscribe, e-mail: [email protected]
For queries about this service, please contact Infrastructure at:
[email protected]
