use regular configure instead of wrapper script for recent UCX easyconfigs

[bedroge]

The UCX easyconfigs for version 1.9.0 and newer are currently using configure_cmd = contrib/configure-release; this is a wrapper script around ./configure which passed some predefined flags (these haven't changed for many years), see:
https://github.com/openucx/ucx/blob/master/contrib/configure-release

While trying to install UCX on RISC-V, I ran into issues with the outdated config.guess included with UCX. The ConfigureMake easyblock should download a newer version and use that one, but it turns out that it only does that when the configure_cmd script includes a certain string (Generated by GNU Autoconf). Though the configure script does include it, this is not the case for the contrib/configure-release wrapper script, hence that step is skipped. By appending the options that this wrapper uses to configopts, we can just call ./configure itself, and then EB does make sure that an updated config.guess is being used. Additionally, it's more explicit in the easyconfig which flags are being passed.

[bedroge]

@boegelbot please test @ generoso

[boegelbot]

@bedroge: Request for testing this PR well received on login1

PR test command 'EB_PR=20428 EB_ARGS= EB_CONTAINER= EB_REPO=easybuild-easyconfigs /opt/software/slurm/bin/sbatch --job-name test_PR_20428 --ntasks=4 ~/boegelbot/eb_from_pr_upload_generoso.sh' executed!

• exit code: 0
• output:

Submitted batch job 13364

Test results coming soon (I hope)...

- notification for comment with ID 2079607477 processed

Message to humans: this is just bookkeeping information for me,
it is of no use to you (unless you think I have a bug, which I don't).

[bedroge]

@boegelbot please test @ jsc-zen3

[boegelbot]

@bedroge: Request for testing this PR well received on jsczen3l1.int.jsc-zen3.fz-juelich.de

PR test command 'if [[ develop != 'develop' ]]; then EB_BRANCH=develop ./easybuild_develop.sh 2> /dev/null 1>&2; EB_PREFIX=/home/boegelbot/easybuild/develop source init_env_easybuild_develop.sh; fi; EB_PR=20428 EB_ARGS= EB_CONTAINER= EB_REPO=easybuild-easyconfigs EB_BRANCH=develop /opt/software/slurm/bin/sbatch --job-name test_PR_20428 --ntasks=8 ~/boegelbot/eb_from_pr_upload_jsc-zen3.sh' executed!

• exit code: 0
• output:

Submitted batch job 4036

Test results coming soon (I hope)...

- notification for comment with ID 2079612506 processed

Message to humans: this is just bookkeeping information for me,
it is of no use to you (unless you think I have a bug, which I don't).

[boegelbot]

Test report by @boegelbot
SUCCESS
Build succeeded for 9 out of 9 (9 easyconfigs in total)
cns1 - Linux Rocky Linux 8.9, x86_64, Intel(R) Xeon(R) CPU E5-2667 v3 @ 3.20GHz (haswell), Python 3.6.8
See https://gist.github.com/boegelbot/755c932646bac0d833b12738b981ac26 for a full test report.

[boegelbot]

Test report by @boegelbot
SUCCESS
Build succeeded for 9 out of 9 (9 easyconfigs in total)
jsczen3c1.int.jsc-zen3.fz-juelich.de - Linux Rocky Linux 9.3, x86_64, AMD EPYC-Milan Processor (zen3), Python 3.9.18
See https://gist.github.com/boegelbot/c47e7ca8c6ba8d9e8e6a90fc030c2709 for a full test report.

[branfosj]

We also use contrib/configure-release in ucx_plugins.py. Do we want to look at changing that as well?

[bedroge]

We also use contrib/configure-release in ucx_plugins.py. Do we want to look at changing that as well?

Ah, good catch, didn't know it was used in other places. I suppose it makes sense to use the same approach everywhere, so I can make a PR for that easyblock as well.

However, @SebastianAchilles found out that this patch is actually not really required anymore to solve the RISC-V issue (since the autoreconf command added in #20429 apparently updates the included config.guess, and then it does recognize RISC-V), so in that sense we don't really need this PR anymore. On the other hand, this approach may still be a bit cleaner anyway, since we don't depend on that wrapper script anymore?

edit: and it may still be useful for UCX 1.16.0. Even though that version does support RISC-V out of the box (and hence doesn't need the patch from #20429), they still ship a very old config.guess (which doesn't detect RISC-V...) in their tarball. This means that this version will either need to do an autoreconf on RISC-V or use the regular configure script (to make sure that EB's config.guess is used).
I can verify this by making a PR for UCX 1.16.0 as well.

[bedroge]

I've applied the same change to the UCX plugins easyblock in easybuilders/easybuild-easyblocks#3315. I've also opened a PR for UCX 1.16.0 (#20431), which demonstrates that this change is also still useful for RISC-V builds.

[SebastianAchilles]

this approach may still be a bit cleaner anyway

Yes, I share your opinion, using the regular configure seems a bit cleaner.

I've applied the same change to the UCX plugins easyblock in easybuilders/easybuild-easyblocks#3315. I've also opened a PR for UCX 1.16.0 (#20431), which demonstrates that this change is also still useful for RISC-V builds.

Since this change will be useful for future UCX easyconfigs for RISC-V I think it makes sense to merge this PR.

@branfosj Do you agree?

[boegel]

make sense to circumvent the contrib/configure-release script, since it doesn't add much value, and causes trouble since updated config.guess is not picked up

[boegel]

Test report by @boegel
SUCCESS
Build succeeded for 9 out of 9 (9 easyconfigs in total)
node3147.skitty.os - Linux RHEL 8.8, x86_64, Intel(R) Xeon(R) Gold 6140 CPU @ 2.30GHz, Python 3.6.8
See https://gist.github.com/boegel/17a637b6b2616e41c6cafbcf3d01f8e9 for a full test report.

[boegel]

Going in, thanks @bedroge!