Fix carry-over only due to one matching bugref in step title

[Martchus]

• Fix carry-over only due to one matching bugref in step title
  • Prevent carry-over despite a mismatch of failing test modules like in
    https://progress.opensuse.org/issues/94880
  • Still allow carry-over when the same bug is found via different modules
    (ef5f07c) but only if the other failing
    test modules match as well
• Consider all modules on carry over, despite missing/broken results
  • If it isn't possible to read the results/details of some (soft)failed
    modules that doesn't mean we should ignore them completely. In addition,
    further (soft)failed modules should not be ignored as well.

---

I've been adding tests now. I've also been testing that tests fail without the change (as there was already a similar test in place which wasn't good enough).

[okurz]

I like the code style change :)

[codecov]

Codecov Report

Merging #3988 (a8b56d1) into master (65cd968) will decrease coverage by 0.03%.
The diff coverage is 100.00%.

@@            Coverage Diff             @@
##           master    #3988      +/-   ##
==========================================
- Coverage   96.93%   96.90%   -0.04%     
==========================================
  Files         370      370              
  Lines       32968    32971       +3     
==========================================
- Hits        31959    31949      -10     
- Misses       1009     1022      +13     

Impacted Files	Coverage Δ
lib/OpenQA/Schema/Result/Jobs.pm	98.50% <100.00%> (-0.01%)	⬇️
t/17-labels_carry_over.t	100.00% <100.00%> (ø)
t/lib/OpenQA/Test/Utils.pm	78.37% <0.00%> (-3.66%)	⬇️

---

Continue to review full report at Codecov.

Legend - Click here to learn more
Δ = absolute <relative> (impact), ø = not affected, ? = missing data
Powered by Codecov. Last update 65cd968...a8b56d1. Read the comment docs.

[Vogtinator]

The algorithm seems sound to me.

The only weakness I could identify is that when a failed module gains a bugref in test code, that blocks carryover. Shouldn't really be an issue in practice, and the previous code had that as well.

[Martchus]

The only weakness I could identify is that when a failed module gains a bugref in test code, that blocks carryover. Shouldn't really be an issue in practice, and the previous code had that as well.

I thought about it as well but thought it wasn't worth it as it would complicate the code a bit. We can still implement it later if needed.

[okurz]

Sorry, I don't understand. What has the "carry-over" have to do with "matching bugref in step title"? I thought carry-over is only looking at failed modules, not steps?

[Martchus]

I thought carry-over is only looking at failed modules, not steps?

That is wrong. With ef5f07c (referenced in the commit message) it has been changed to also take bugrefs within step titles into account.