Pull request CI report

Report generated at: 2024-07-08 03:07 UTC
Newest commit scanned: 3355c6a
Status: ✅ good

There are existing issues already. Please look into the report to make sure none of them affect the packages in question:
https://qa-reports.gentoo.org/output/gentoo-ci/f8d1b9821e/output.html

Pull request CI report

Report generated at: 2024-07-09 00:07 UTC
Newest commit scanned: d6b7459
Status: ✅ good

There are existing issues already. Please look into the report to make sure none of them affect the packages in question:
https://qa-reports.gentoo.org/output/gentoo-ci/f006124ff8/output.html

Pull request CI report

Report generated at: 2024-07-10 00:02 UTC
Newest commit scanned: 6bac88e
Status: ✅ good

There are existing issues already. Please look into the report to make sure none of them affect the packages in question:
https://qa-reports.gentoo.org/output/gentoo-ci/d4198251a6/output.html

Pull request CI report

Report generated at: 2024-07-10 10:22 UTC
Newest commit scanned: a1a11b2
Status: ✅ good

There are existing issues already. Please look into the report to make sure none of them affect the packages in question:
https://qa-reports.gentoo.org/output/gentoo-ci/4543f8d5c6/output.html

Pull request CI report

Report generated at: 2024-07-11 00:42 UTC
Newest commit scanned: 14efa54
Status: ✅ good

There are existing issues already. Please look into the report to make sure none of them affect the packages in question:
https://qa-reports.gentoo.org/output/gentoo-ci/4b1fef4d3d/output.html

wrt java see Required Java dependencies for optional Java support

Thank you for the review, @vaukai!

So far I've updated the ebuild in my overlay.
Do you mind if I add Co-authored-by and Signed-off-by to the commit message here?

Co-authored-by: Volkmar W. Pogatzki <gentoo@pogatzki.net>
Signed-off-by: Volkmar W. Pogatzki <gentoo@pogatzki.net>

Thank you for the review, @vaukai!

So far I've updated the ebuild in my overlay. Do you mind if I add Co-authored-by and Signed-off-by to the commit message here?

Co-authored-by: Volkmar W. Pogatzki <gentoo@pogatzki.net>
Signed-off-by: Volkmar W. Pogatzki <gentoo@pogatzki.net>

Please don't since that's unusual in Gentoo. Adjusting the ebuild according to Java_Developer_Guide is enough.

If you like you could try getting rid of java-ant-2 like i did this morning for mythtv but that's not a trivial change. I would otherwise do that after your PR got merged.
The reason for dropping java-ant-2 is that it runs BSFIX aka xml-rewrite provided by an undocumented black box called javatoolkit.

meanwhile i have 4.9.0-r2 ready to merge which no longer inherits java-ant-2.
Please try including those changes here.

@Jamim you want to move changelog from commit message to metadata.xml (compare https://gitweb.gentoo.org/repo/gentoo.git/tree/dev-java/tomcat-native/metadata.xml)

Pull request CI report

Report generated at: 2024-07-11 14:07 UTC
Newest commit scanned: de32b9f
Status: ✅ good

There are existing issues already. Please look into the report to make sure none of them affect the packages in question:
https://qa-reports.gentoo.org/output/gentoo-ci/4ed1ae1fed/output.html

@vaukai I've added changelog and doc links to the metadata.xml.
However, I prefer to keep a link to the certain changelog section in the commit message.

Pull request CI report

Report generated at: 2024-07-11 14:32 UTC
Newest commit scanned: 5249f21
Status: ✅ good

There are existing issues already. Please look into the report to make sure none of them affect the packages in question:
https://qa-reports.gentoo.org/output/gentoo-ci/1fb99a6e32/output.html

However, I prefer to keep a link to the certain changelog section in the commit message.

Please don't...

Hello @eli-schwartz,
Thank you! I've split the PR into distinct commits.

Hello @negril,

Could you please review this PR?

Thanks in advance!

14efa54 builds fine for me with nVIDIA proprietary drivers.

All branches after that fail config with the following error:

 * abi_x86_64.amd64: running multilib-minimal_abi_src_configure
__nvcc_device_query failed to call cudaLoader::cuInit(0) with error 0x64 (CUDA_ERROR_NO_DEVICE)
__nvcc_device_query failed to call cudaLoader::cuInit(0) with error 0x64 (CUDA_ERROR_NO_DEVICE)
 * ERROR: media-libs/opencv-4.10.0::local_overlay failed (configure phase):
 *   Failed to get CUDA host native arch
 * 
 * Call stack:
 *     ebuild.sh, line  136:  Called src_configure
 *   environment, line 5698:  Called cmake-multilib_src_configure
 *   environment, line 1335:  Called multilib-minimal_src_configure
 *   environment, line 4407:  Called multilib_foreach_abi 'multilib-minimal_abi_src_configure'
 *   environment, line 4657:  Called multibuild_foreach_variant '_multilib_multibuild_wrapper' 'multilib-minimal_abi_src_configure'
 *   environment, line 4362:  Called _multibuild_run '_multilib_multibuild_wrapper' 'multilib-minimal_abi_src_configure'
 *   environment, line 4360:  Called _multilib_multibuild_wrapper 'multilib-minimal_abi_src_configure'
 *   environment, line  608:  Called multilib-minimal_abi_src_configure
 *   environment, line 4401:  Called multilib_src_configure
 *   environment, line 4923:  Called cuda_get_host_native_arch
 *   environment, line 1814:  Called die
 * The specific snippet of code:
 *       : "${CUDAARCHS:=$(__nvcc_device_query || __nvcc_device_query || die 'Failed to get CUDA host native arch')}";

For comparison, the 14efa54 successful config looks like so:

 * abi_x86_64.amd64: running multilib-minimal_abi_src_configure
__nvcc_device_query failed to call cudaLoader::cuInit(0) with error 0x64 (CUDA_ERROR_NO_DEVICE)
 * CUDAHOSTCXX: gcc-13
 * CUDAARCHS: 
 * CUDA_ARCH_BIN: 5.2
 * CUDA_ARCH_PTX: 

Hello @Delicates,

First of all, thank you for testing! 🙇🏼

Normally, __nvcc_device_query just prints 50 on my laptop with GeForce GTX 960M:

$ __nvcc_device_query
50

But in rare cases, it fails with the following message:

__nvcc_device_query failed to call cudaLoader::cuInit(0) with error 0x3e7 (CUDA_ERROR_UNKNOWN)

With 14efa54, __nvcc_device_query's failures were ignored, so that's why CUDAARCHS was empty. And in my case, empty CUDAARCHS value caused CUDA_ARCH_BIN/CUDA_ARCH_PTX miscalculation in OpenCV's CMake scripts, so I had 52 instead of 50 which was not really compatible with my GPU. That's why I added a retry and the explicit die when it fails twice in a row.

In your case, I see that __nvcc_device_query constantly can't find any CUDA device for some reason.
Can you please provide a bit more information about your system?

• what GPUs are installed?
• what nvidia-smi shows?

Also, it shouldn't fail in case you set a desired value to CUDAARCHS in make.conf or to the environment when you build opencv.

I've get rid of the die in favor of a warning for now. You can test it with 83a0530.

Also, I've found a very annoying bug related to CUDA configuration:

• 🐛 Incompatibility of CUDA CMake options opencv/opencv#25920

I'm not sure what's the best way to handle this situation, so any suggestions are welcome.

Thanks!

Pull request CI report

Report generated at: 2024-07-17 04:38 UTC
Newest commit scanned: 83a0530
Status: ✅ good

There are existing issues already. Please look into the report to make sure none of them affect the packages in question:
https://qa-reports.gentoo.org/output/gentoo-ci/e6c9ae5966/output.html

Прывітанне @Jamim

Thanks for tackling this OpenCV version bump, it's been a blocker for many things.

I've get rid of the die in favor of a warning for now. You can test it with 83a0530.

This seems to work better than any of the prior versions without any make.conf updates - no config error at all:

 * abi_x86_64.amd64: running multilib-minimal_abi_src_configure
 * CUDAHOSTCXX: gcc-13
 * CUDAARCHS: 52
 * CUDA_ARCH_BIN: 5.2
 * CUDA_ARCH_PTX: 52
 * CUDA_GENERATION: Maxwell

With 14efa54, __nvcc_device_query's failures were ignored, so that's why CUDAARCHS was empty. And in my case, empty CUDAARCHS value caused CUDA_ARCH_BIN/CUDA_ARCH_PTX miscalculation in OpenCV's CMake scripts, so I had 52 instead of 50 which was not really compatible with my GPU. That's why I added a retry and the explicit die when it fails twice in a row.

I know 52 is currently the default arch in CUDA, and it would be annoying if it doesn't match your GPU.
My GPU actually happens to be 52 arch, so I guess the default always works in my favour.

+-----------------------------------------------------------------------------------------+
| NVIDIA-SMI 555.58.02              Driver Version: 555.58.02      CUDA Version: 12.5     |
|-----------------------------------------+------------------------+----------------------+
| GPU  Name                 Persistence-M | Bus-Id          Disp.A | Volatile Uncorr. ECC |
| Fan  Temp   Perf          Pwr:Usage/Cap |           Memory-Usage | GPU-Util  Compute M. |
|                                         |                        |               MIG M. |
|=========================================+========================+======================|
|   0  NVIDIA GeForce GTX 970         Off |   00000000:02:00.0  On |                  N/A |
| 42%   40C    P0             56W /  250W |    1500MiB /   4096MiB |      8%      Default |
|                                         |                        |                  N/A |
+-----------------------------------------+------------------------+----------------------+

I did already have a few variables defined in my make.conf (not CUDAARCHS though), but looks like they were not helping with the previous version errors:

NVCCFLAGS="-O2 -gencode arch=compute_52,code=sm_52 -allow-unsupported-compiler"
CUDA_NVCC_FLAGS="${NVCCFLAGS}"
NVCC_FLAGS_EXTRA="${NVCCFLAGS}"
NVCC_PREPEND_FLAGS="${NVCCFLAGS}"
LLVM_TARGETS="NVPTX X86"

CUDA_NVCC_FLAGS="${NVCCFLAGS}"
CUDA_ARCH="sm_52"
CUDA_ARCH_BIN="5.2"
CUDA_GENERATION="Maxwell"
TORCH_CUDA_ARCH_LIST="5.2"

Hello @vaukai,
I've ported your changes from #37823 to this PR.
Thank you!

Pull request CI report

Report generated at: 2024-09-11 10:24 UTC
Newest commit scanned: c0b97f4
Status: ✅ good

There are existing issues already. Please look into the report to make sure none of them affect the packages in question:
https://qa-reports.gentoo.org/output/gentoo-ci/5b0efc1e68/output.html

Hello @eli-schwartz,
Do you still have any concerns here?

Hello @vaukai, I've ported your changes from #37823 to this PR. Thank you!

lgtm

Hello @juippis,

You asked me to stop pinging people and suggested to take a break.
Well, as you can see, this approach failed completely.

Does this PR have chances to be merged this year or, at least, to receive some additional feedback and be improved while I don't ping anyone?
I don't think so.

I state that there's something fundamentally wrong with contribution processing in Gentoo, so not only this PR got stuck, but hundreds of other PRs, whether they are good or bad, are lost and become irrelevant over time.

It's clear that there are not enough resources to process all the PRs, so maybe it's time to stop processing them at all for a moment, take a deep breath and think about what can be changed in the workflow? I believe there is room for improvements and optimizations, so maintainers can bring more value while spend less time. Please let me know if I can do anything to make it happen.

For context, I use Gentoo Linux since 2007.
Only a year ago or so I started trying to do my best to not only receive from it but also to return something valuable back. And what I can tell you so far is I was happier about Gentoo when I didn't care and just used it.

I understand this is not the right place to start this topic, but I'm tired and quite disappointed so here we go.

Thank you for your time and patience in reading this!

P.S. It would be nice if you share my concerns with those whom I shouldn't ping there.

@Jamim,

The third PR still contains many changes that are all grouped into a single commit, with the implication that these are issues you encountered in 4.9.0-r2 as well. As I said above:

• in general, if your commit message says "also changes" with a list of "also changed" things, it may be a sign that the things you are changing have nothing to do with the subject of the commit, and could be done via a separate commit

My initial review comment therefore is still outstanding.

And e.g. for https://bugs.gentoo.org/930368 you are NOT in fact fixing it, sorry -- so please remove the "Closes" link entirely if you are not going to do it correctly. (This is, again, the issue with making many changes in the same commit as incrementing the version number, when the changes aren't a result of incrementing the version number.)

And changes such as depending on a modified acct-user/portage package to "Allow access to GPUs during tests run" is a very... interesting change. I told you to divide it up into separate commits as a matter of git hygiene, that's not the same as saying I agree with such an acct-user change. But that aside, you haven't followed the bot instructions to

In order to force reassignment and/or bug reference scan, please append [please reassign] to the pull request title.

which appears to be necessary in order for the bot to notice that you are modifying the acct-user/portage package so the maintainers of that package are even notified of the need to review this PR.

The proxied maintainer for opencv itself still hasn't reviewed the PR, but that may be because it was hard to figure out what it's trying to do considering that it lists 13 different change requests in the commit message for the third commit and it's not necessarily clear which changes are tied to which change requests.

Pull Request assignment

Submitter: @Jamim
Areas affected: ebuilds
Packages affected: acct-user/portage, media-libs/opencv

acct-user/portage: @gentoo/portage
media-libs/opencv: @negril, @gentoo/proxy-maint

Linked bugs

Bugs linked: 930368, 930682, 928747

In order to force reassignment and/or bug reference scan, please append [please reassign] to the pull request title.

Docs: Code of Conduct ● Copyright policy (expl.) ● Devmanual ● GitHub PRs ● Proxy-maint guide

Pull request CI report

Report generated at: 2024-11-02 01:33 UTC
Newest commit scanned: 5fb47eb
Status: ✅ good

There are existing issues already. Please look into the report to make sure none of them affect the packages in question:
https://qa-reports.gentoo.org/output/gentoo-ci/4cf33ce414/output.html

Hello @eli-schwartz,

The third PR still contains many changes that are all grouped into a single commit, with the implication that these are issues you encountered in 4.9.0-r2 as well. As I said above:

• in general, if your commit message says "also changes" with a list of "also changed" things, it may be a sign that the things you are changing have nothing to do with the subject of the commit, and could be done via a separate commit

It's not also changes now. It's changes in comparison to. As far as I can understand, you want me to copy 4.9.0-r2 as is to 4.10.0 in a dedicated commit and then to commit all the changes one by one, separately. If so, it would be perfectly fine to split the commit if I wanted to modify 4.9.0-r2. But it's not the case. opencv-4.10.0 is a new ebuild, and I don't want to add unnecessary commits here. No one except you asks for such a detailed history, so I don't see any benefits of doing so.

And e.g. for bugs.gentoo.org/930368 you are NOT in fact fixing it, sorry -- so please remove the "Closes" link entirely if you are not going to do it correctly. (This is, again, the issue with making many changes in the same commit as incrementing the version number, when the changes aren't a result of incrementing the version number.)

I suppose you got something mixed up. There was no Closes link in commit messages. There were Bug links only. So there's nothing to remove. In the meantime, any suggestions on how to correctly resolve 930368 are welcome.

And changes such as depending on a modified acct-user/portage package to "Allow access to GPUs during tests run" is a very... interesting change. I told you to divide it up into separate commits as a matter of git hygiene, that's not the same as saying I agree with such an acct-user change.

Do you have any better ideas on how to allow tests to use the GPU? Is there any... boring way to do so?

But that aside, you haven't followed the bot instructions to

In order to force reassignment and/or bug reference scan, please append [please reassign] to the pull request title.

which appears to be necessary in order for the bot to notice that you are modifying the acct-user/portage package so the maintainers of that package are even notified of the need to review this PR.

Thanks for catching this! 🙇🏼

The proxied maintainer for opencv itself still hasn't reviewed the PR, but that may be because it was hard to figure out what it's trying to do considering that it lists 13 different change requests in the commit message for the third commit and it's not necessarily clear which changes are tied to which change requests.

Sorry, but it looks like a pure speculation. I believe @negril can speak for himself if he wants to.

Current opencv version in Gentoo is 4.12.0-r1
Maybe this PR could be closed or rebased with the video useflag change?