Issue with semantic cache tests

[alkidbaci]

I noticed that after the changes on PR #512 the GitHub Actions will sometime fail and sometime pass.
The error that makes it to fail is the same as one which has occurred in a previous commit of PR #512:

/home/runner/work/_temp/8f841887-af9a-4a6e-9efc-95f025472c96.sh: line 6:  2208 Segmentation fault      (core dumped) pytest -p no:warnings -x

Sometime it fails during coverage report (which re-runs the tests) but the error is similar.

/home/runner/work/_temp/8c2f5eb6-167a-4810-9be6-bbb38a3ecc26.sh: line 2: 244567 Segmentation fault      (core dumped) coverage run -m pytest

@Louis-Mozart any idea what may be causing the issue? Is there anything of a "random" nature in your code?
If you cant find the cause maybe you can narrow down which specific part of the test at tests/test_semantic_cache may cause it so that we can comment it out for the moment, until we find a solution.

[Louis-Mozart]

Yes, I think I know where the issue comes from and it is now solved. If you check the commits where the runs fail, I do not provide enough size for the cache algorithm (see line 10) and even though the tests were working locally, it was not working on GitHub so when I allocated more than enough memory for the cache, all the tests worked successfully. So I think the problem was a memory issue and there is no randomness in the code. If you rerun the tests now, they will always work.

[alkidbaci]

Okay, maybe I did not state the issue clearly. Before I merged your PR #512 the tests had passed indeed. After your changes were now in the develop branch the error keeps showing up when I try to merge other pull requests. For example check this report here.
I said "random" because sometime such error will occur and sometimes it does not. The problem is apparently not fixed. I have to re-run the GitHub action workflow sometimes for the tests to pass but that is not something we want to do in practice. The GitHub action workflow should run only once and if everything is supposed to be ok then everything should pass (there should be no need to re-run it). Maybe the "randomness" its not in your code but is related with the tests executed earlier and the memory loaded more than usual by the time it comes to your test.

@Louis-Mozart if you can do something about this, please do so, otherwise I will take this over (which will cost me some time because I will have to understand your code first).

I also noticed that sometimes the workflow never terminates and it automatically stops after crossing the 6h runtime restriction. When I had a closer look, I think the cause of this non-terminatable execution is test_roces.py.

I would appreciate your input on this issue @Jean-KOUAGOU.

[Louis-Mozart]

I am afraid there is no much I can do as I don't have any idea of why this is happening. I have been running all the tests locally several times and it always works.

[Jean-KOUAGOU]

Maybe we can skip the semantic cache test for the moment. Then, we can make a new release?

…

On Mon, Mar 3, 2025, 13:45 Louis-Mozart ***@***.***> wrote: I am afraid there is no much I can do as I don't have any idea of why this is happening. I have been running all the tests locally several times and it always works. — Reply to this email directly, view it on GitHub <#515 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/AK4X4EV3PANBRWAP65QJ3732SRFE3AVCNFSM6AAAAABX2GXHCWVHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMZDMOJUGI3TMOBWGI> . You are receiving this because you were mentioned.Message ID: ***@***.***> [image: Louis-Mozart]*Louis-Mozart* left a comment (dice-group/Ontolearn#515) <#515 (comment)> I am afraid there is no much I can do as I don't have any idea of why this is happening. I have been running all the tests locally several times and it always works. — Reply to this email directly, view it on GitHub <#515 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/AK4X4EV3PANBRWAP65QJ3732SRFE3AVCNFSM6AAAAABX2GXHCWVHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMZDMOJUGI3TMOBWGI> . You are receiving this because you were mentioned.Message ID: ***@***.***>