Changed wheel name to smlptech

[mdmitry1]

1. Changed wheel name to smlptech-1.0.1-cp311-cp311-manylinux_2_24_x86_64.manylinux_2_28_x86_64.whl
2. Rebuilt all three wheel test images
3. Regression passed for all images

[fbrausse]

I suppose there will be a name conflict with this other smlp package if we keep installing bin/smlp and also with the package name smlp used here

smlp/pyproject.toml

Line 38 in 8f6dc4c

smlp = "smlp.run_smlp:main2"

and in the source files smlp_{optimize,solver,verify,query,terms}.py

[mdmitry1]

I've not seen any conflict, if fresh virtual environment is used.
I run regression and it passed.
In Docker there is also no conflict.

[fbrausse]

You are renaming the wheel due to the name conflict with that other "smlp" package on pypi, yes? As that one provides a Python package called smlp that one can import with import smlp, there is a conflict if we keep our package name smlp as well, is there not?

[mdmitry1]

Conflict will occur only if someone needs both smlp packages simultaneously.
I believe that chances for this are slim

[fbrausse]

Why do we need the *.log files in tests/smlp_regression? I particular, since 36 tests seem to have the status "Failed". I think it is better to not have an official reference file that claims that regression tests fail.

[fbrausse]

Using the two patches mentioned earlier, I am able to build, install and successfully test the smlptech wheel.

However, I would still like to know why this PR includes the *.log files with failed tests. Are they expected to fail?

[mdmitry1]

Hi Franz,

Yes - many tests are expected to fail.
Therefore I'm comparing new failures to the expected.
It can be easily validated by running regression on any of previous tags.
Now we have more failed tests then before as after using pip install order of the attributes in the json file has changed in some tests.
As soon as regression will be cleaned, this won't be necessary.
If I'm not mistaken, regression is supposed to be cleaned before the release.

Dmitry

[fbrausse]

Yes - many tests are expected to fail.

I could imagine users trying out SMLP to be surprised by failing tests. Should we document somewhere (in some README, I guess) the reasons for failing tests and which tests are expected to fail?

Now we have more failed tests then before as after using pip install order of the attributes in the json file has changed in some tests.

Indeed, I've also seen this phenomenon. Not sure where it comes from, but for me it's clear: order shouldn't matter in those JSON objects. Similar to many of the lines of log messages: they are not relevant to the exact test result and should be ignored. Some are ignored by the smlp_regr.py script, but that needs to change to a more opt-in approach in my opinion. Line-by-line compatibility is not what we're aiming for with SMLP.

As soon as regression will be cleaned, this won't be necessary.
If I'm not mistaken, regression is supposed to be cleaned before the release.

Maybe I missed this before, what does "cleanup of regression tests" mean? What are the plans here?

[zurabksmlp]

Likely many of the failures are due to missing test input data? I mean the recovered tests -- recovered missing input data and spec files. <https://www.avast.com/sig-email?utm_medium=email&utm_source=link&utm_campaign=sig-email&utm_content=webmail> Virus-free.www.avast.com <https://www.avast.com/sig-email?utm_medium=email&utm_source=link&utm_campaign=sig-email&utm_content=webmail> <#DAB4FAD8-2DD7-40BB-A1B8-4E2AA1F9FDF2>

…

On Thu, Mar 26, 2026 at 9:39 PM Franz Brauße ***@***.***> wrote: *fbrausse* left a comment (SMLP-Systems/smlp#74) <#74?email_source=notifications&email_token=BPWPDLYBOYQHFQSXMYTJYUL4SWBQNA5CNFSNUABFM5UWIORPF5TWS5BNNB2WEL2JONZXKZKDN5WW2ZLOOQXTIMJTG43TEMJSGYY2M4TFMFZW63VGMFZXG2LHN2SWK5TFNZ2LK4DSL5RW63LNMVXHIX3POBSW4X3DNRUWG2Y#issuecomment-4137721261> Yes - many tests are expected to fail. I could imagine users trying out SMLP to be surprised by failing tests. Should we document somewhere (in some README, I guess) the reasons for failing tests and which tests are expected to fail? Now we have more failed tests then before as after using pip install order of the attributes in the json file has changed in some tests. Indeed, I've also seen this phenomenon. Not sure where it comes from, but for me it's clear: order shouldn't matter in those JSON objects. Similar to many of the lines of log messages: they are not relevant to the exact test result and should be ignored. Some are by the smlp_regr.py script, but that needs to change to a more opt-in approach in my opinion. Line-by-line compatibility is not what we're aiming for with SMLP. As soon as regression will be cleaned, this won't be necessary. If I'm not mistaken, regression is supposed to be cleaned before the release. Maybe I missed this before, what does "cleanup of regression tests" mean? What are the plans here? — Reply to this email directly, view it on GitHub <#74?email_source=notifications&email_token=BPWPDLYBOYQHFQSXMYTJYUL4SWBQNA5CNFSNUABFM5UWIORPF5TWS5BNNB2WEL2JONZXKZKDN5WW2ZLOOQXTIMJTG43TEMJSGYY2M4TFMFZW63VGMFZXG2LHN2SWK5TFNZ2LK4DSL5RW63LNMVXHIX3POBSW4X3DNRUWG2Y#issuecomment-4137721261>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/BPWPDL2YY65SJ6DYUHEFMVT4SWBQNAVCNFSM6AAAAACW7EGUACVHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHM2DCMZXG4ZDCMRWGE> . You are receiving this because you were assigned.Message ID: ***@***.***>

[mdmitry1]

Cleanup of regression tests mean that regression for the released version will pass

[fbrausse]

Cleanup of regression tests mean that regression for the released version will pass

Short of disabling those tests I don't see how we'll manage that. We can discuss tomorrow or later. For this PR, as we already have some of those logs somehow, it's fine with me.