[v7r2] Move to a src/ layout#4858
Conversation
| # gracefully, make them non-daemonic and use a suitable | ||
| # signalling mechanism such as an Event." | ||
| - pytest --no-cov -k 'not test_BaseType_Unicode and not test_nestedStructure and not testLockedClass' | ||
| # - pytest --no-cov src/DIRAC/Core/Security/test |
There was a problem hiding this comment.
@chaen In the previous Python 3 merge request I had missed this test. I tried adding it here but it fails in a non obvious way. Would you be able to take a look at some point?
(if you can you should work on top of this branch as I fixed all except one of the failures already)
There was a problem hiding this comment.
Ah you mean the Security tests ? Yes, these are nasty, because they unload some modules from memory in order to do a pyGSI/M2Crypto comparison. I think they can be simplified, and then not run anymore as a specific tests. I'll have a look !
There was a problem hiding this comment.
So that seems to be an M2Crypto issue. I have opened a bug report. I suspect that the problem lies in the C-API, probably some mess with str/bytes but since I have no knowledge of that, I will wait to see if there is a reaction before embarking in fixing it myself
|
I have seen you moved tests/Utilities → src/DIRAC/tests/Utilities/ |
Unfortunately it's a bit messy as some bits of test code are being imported as: from DIRAC.tests.Utilities import somethingThere are other ways of fixing this this and I'm happy to do it differently if you have a strong opinion? I think the correct solution is probably to tidy up the tests a bit but that's a task for another PR. |
|
Alright. #4859 |
|
There is a lot of documentation out there that points to the location of dirac-install.py where it is at the moment in the integration branch on github, isn't there? |
|
Excellent observation @andresailer I re-launch: what if we move dirac-install.py in management repo? It has always been a script not depending from any specific release, and, in a way, this PR is also a first step towards retirement of dirac-install script. |
|
(In the current model of packaging and updating) Is there a use case for |
|
dirac-install is used to upgrade the already existing installation. In this case typically it is taken from the installation itself. However, it is the same dirac-install (may be a different version) as the one for an initial installation. So, moving it to management does not harm |
Won't moving it to |
|
Well, we can replace the dirac-install script that's here with a script that just downloads the dirac-install that is in management and run it. This would BTW make sure that we don't run into backward incompatibilities. |
|
Is the dirac-distribution.py and the related stuff updated in the management project to take this change into account ? |
|
This change is probably not so complex but still can have lots of repercussions here and there. So, I would suggest to make it the first one in v7r3 release. I mean I will make rel-v7r2 branch with the current integration contents which should be just certified for the release. Then I merge this one into integration straight away. What do you think ? |
|
I think that this PR can be merged as it is in v7r2. For the move of dirac-install out of this repo we can wait v7r3. |
|
Can we please keep a copy of dirac-install at its current location, and not move it a week before the end of year break? Maybe add a deprecation print out at the end of the script? Ideally already putting one in diracgrid/management and referring to that location in the previously mentioned deprecation message? |
|
I opened issue #4860 to discuss the "dirac-install move" topic, as that won't be part of this PR. Here, what should be addressed is only this:
|
I would really like to include this and the
I agree, I'll add a copy of it now. |
|
I guess I will not be able to compile the v7r2 distribution tar balls if this PR is merged ? |
@atsareg Can you point me to the docs on exactly how you do this so I can check? |
|
Basically: you would need to modify a line or 2, probably here: https://github.com/DIRACGrid/management/blob/master/dirac-distribution/dirac-create-distribution-tarball.py#L463 (I only had a very quick look) and then the docker container will need to be re-created. To force that, temporary change this line: https://github.com/DIRACGrid/management/blob/master/.github/workflows/images-creator.yml#L14 |
|
I've just opened the PR: DIRACGrid/management#28 I've tested it locally with a hack to make it clone this branch instead of the actual tag and it seems to work. |
|
Yes, merged. The containers are created here: https://github.com/DIRACGrid/management/actions/runs/410422572 |
|
I've checked and the new image works as expected. @atsareg Can you ping me once you've uploaded the tarballs so I can run some checks before tomorrow's hackathon rather than risking it needing to be cancelled |
The first step in implementing what was discussed at the last BiLD. Obviously this PR touches a lot of files but it should be quite easy to review as there are very few actual changes. It is split into three commits:
First part of #4774
BEGINRELEASENOTES
*Python 3
NEW: Moved to a src/ repository layout
ENDRELEASENOTES