@TimoWilken The CI errors seems unrelated, or am I mistaken?

@jgrosseo the libjalienO2 dependency (added in this PR) isn't picked up by QC. I'll open a PR in alidist and restart these checks when that's merged. While the O2 build runs fine, could you hold off on merging for now so we don't break the QC build?

QC should not need to be touched. Dependencies should be transitive in both aliBuild and CMake.

I don't have a solution but this is not going to work:
HINTS "${LIBJALIENO2}"

This variable is unknown in downstream packages. You could use the environment variable LIBJALIENO2_ROOT

@costing Do you know who could follow this up after @Atlantic777 departure. My understanding is that just the cmake has to be fixed.

@jgrosseo that would probably be me. I read through this thread but I'm not entirely sure what is going on. If you know what needs to be fixed in cmake please let me know and I will have a look.

Apart from resolving the merge conflict which is probably simple, follow the comments by @Barthelemy about the different variable name

Thanks for looking into it!

Hi @Barthelemy @jgrosseo @iraklic,
Isn't ${LIBJALIENO2} passed to cmake in o2 alidist script? Please correct me if I am wrong.

@yuw726 the cmake variable LIBJALIENO2 is set to the value of the environment variable LIBJALIENO2_ROOT when configuring O2.

Later in the build, QualityControl is built and when it loads the dependencies it tries to use the FindlibjalienO2.cmake and fails because, at this stage, LIBJALIENO2 is not knows. You should try to use the environment variable LIBJALIENO2_ROOT in the Find script. Not guaranteed to work but it is worth trying.

To test do aliBuild build QualityControl.

@iraklic Any news?

@iraklic Any news?

Sorry, I am having very busy week. It is on my todo. BTW: do I understand it right that you expect a new PR from me with the conflict in CCDB/src/CcdbApi.cxx resolved or there is some other way of modifying this PR (I have never done that)?

@iraklic Any news?

Sorry, I am having very busy week. It is on my todo. BTW: do I understand it right that you expect a new PR from me with the conflict in CCDB/src/CcdbApi.cxx resolved or there is some other way of modifying this PR (I have never done that)?

You cannot push to the branch which made this PR, but you can just start from that branch, push that into your own fork, and then open a PR based on that one. This will include all these changes. We will at that point close this PR.

@iraklic Any news?

Sorry, I am having very busy week. It is on my todo. BTW: do I understand it right that you expect a new PR from me with the conflict in CCDB/src/CcdbApi.cxx resolved or there is some other way of modifying this PR (I have never done that)?

You cannot push to the branch which made this PR, but you can just start from that branch, push that into your own fork, and then open a PR based on that one. This will include all these changes. We will at that point close this PR.

Thank you. That's what I was going to do, but wanted to confirm.

@iraklic Any news?

Sorry, I am having very busy week. It is on my todo. BTW: do I understand it right that you expect a new PR from me with the conflict in CCDB/src/CcdbApi.cxx resolved or there is some other way of modifying this PR (I have never done that)?

You cannot push to the branch which made this PR, but you can just start from that branch, push that into your own fork, and then open a PR based on that one. This will include all these changes. We will at that point close this PR.

Thank you. That's what I was going to do, but wanted to confirm.

Any news?

Fixed conflict

Done, thank you @TimoWilken ! I think it's also requiring a review from @Barthelemy

@costing could you give me a bit of context ?

@costing could you give me a bit of context ?

It is fine. I approved it.

fullCI is unrelated. I'll ignore the space checker on this one (apologies @vkucera , we'll fix this later), and merge when the other two CIs have passed.

I tried to compile and got the following error:

2021-05-27@11:25:43:DEBUG:QualityControl:QualityControl:0: -- The following REQUIRED packages have not been found:
2021-05-27@11:25:43:DEBUG:QualityControl:QualityControl:0: 
2021-05-27@11:25:43:DEBUG:QualityControl:QualityControl:0:  * libjalienO2
2021-05-27@11:25:43:DEBUG:QualityControl:QualityControl:0:    For CCDB API
2021-05-27@11:25:43:DEBUG:QualityControl:QualityControl:0: 

The command I used on CC7 is

aliBuild build QualityControl --defaults o2 --debug

alidist is up to date with master

@TimoWilken do you know why QC misses this?

I am trying to understand nuts and bolts of what is going on here, but I am having trouble to compile dev at this point. It crushes with this:
/home/iraklic/jAliEn_Clean/sw/SOURCES/Monitoring/v3.8.6/v3.8.6/src/Transports/Unix.cxx:18:10: fatal error: filesystem: No such file or directory #include <filesystem> ^~~~~~~~~~~~ compilation terminated.

I checked out dev and building with aliBuild build O2 --defaults o2

Any idea?

some of your stuff is not up to date. Make sure you update O2

What kind of stuff? I pulled O2 using aliBuild init O2@dev --defaults o2 and even did git pull today just in case and tried to rebuild. But have the same issue.

Is there a mattermost/slack channel or something similar when I can ask about my building error? Or should I use alice-project-analysis-task-force mailing list? Or is there a fail-safe tag for O2 I can check out that will surely build? Any help is appreciated.

OK. looks like I am not alone in this: https://alice.its.cern.ch/jira/browse/O2-2327. But the problem is that the workaround yields an entirely new issue:

In file included from /home/iraklic/jAliEn_Clean/sw/SOURCES/O2/master/0/Common/Utils/src/ShmManager.cxx:19:
/home/iraklic/jAliEn_Clean/sw/SOURCES/O2/master/0/Common/Utils/include/CommonUtils/ShmManager.h:40:20: error: field 'counter' has incomplete type 'std::atomic<int>'
   40 |   std::atomic<int> counter = 0; // atomic counter .. counter number of attached processes
      |                    ^~~~~~~

Any suggestions are appreciated

The JIRA ticket you quote is closed and it is said " This should not be the case anymore, provided you move to alibuild 1.8.4 " .
Which alibuild do you have?

You can ask in O2 release integration or O2 analysis challenge on mattermost any additional questions.

This PR did not have any update in the last 30 days. Is it still needed? Unless further action in will be closed in 5 days.

@jgrosseo has this been merged through other PR? I am asking because there are mentions of conflicts mentioned here being fixed and so on. The compilation problem that I had I fixed using the workaround suggested in the ticket that I mentioned. Anyways please let me know if this has been merged though some other PR or if you want me to check it on my own and open new PR with this change. Thanks.

@iraklic I don't think this has been merged through another PR. It is just that other changes had been made to the same code and the conflicts need to be resolved.

I reopened it and do not see a conflicts actually.
However, there are failures in the CI? Could you update this PR with what needs to be done, so that it goes green and we can merge it?

I will try to fix the CMake part and provide modifications in the alidist recipes since the CI issues are related to the discovery of the package.

The following PR alisw/alidist#3275 should solve the issues we observe in the CI.