Improve coverage in multiprocessing cases

[EmmaRenauld]

Our coverage with codecov does not read lines that are called through multiprocessing (Pool).

I separated the case nbr_processes=1 :

if nbr_processes == 1:
    call_it_ourselves()
else:
    as_before

All impacted scripts tested locally with both --processes 1 and --processes 4 (I changed manually the values in each script test).

[codecov]

Codecov Report

Attention: Patch coverage is 44.04762% with 94 lines in your changes missing coverage. Please review.

Project coverage is 69.30%. Comparing base (e7337af) to head (06122c1).
Report is 159 commits behind head on master.

Additional details and impacted files

@@            Coverage Diff             @@
##           master    #1053      +/-   ##
==========================================
+ Coverage   68.93%   69.30%   +0.36%     
==========================================
  Files         438      445       +7     
  Lines       22889    23891    +1002     
  Branches     3106     3262     +156     
==========================================
+ Hits        15779    16557     +778     
- Misses       5793     5958     +165     
- Partials     1317     1376      +59     

Components	Coverage Δ
Scripts	69.96% <ø> (+0.12%)	⬆️
Library	68.40% <44.04%> (+0.74%)	⬆️

[pep8speaks]

Hello @EmmaRenauld, Thank you for updating !

• In the file scilpy/reconst/sh.py:

Line 344:80: E501 line too long (81 > 79 characters)

Comment last updated at 2024-11-21 17:08:12 UTC

[EmmaRenauld]

Not changed: everything in scilpy.tracking.Tracker (too complicated).

I used the keyword search "Pool" to find these methods. There could be others.

[EmmaRenauld]

That gave us a 0.36% upgrade. I was hoping for more, but still.

File by file:
scilpy.reconst.divide: 29% to 69%
scilpy.reconst.fodf: 66% to 71%
scilpy.segment.voting_scheme: 77% to 81%
scilpy.tractanalysis.fixel_density: 52% to 94%

[arnaudbore]

LGTM !

[CHrlS98]

I read the code and everything looks good to me. I approve!

It's sad though that we need to add cases to our code to increase our code coverage by 0.36%. Also for my information, is it only multiprocessing that codecov has a problem handling or is it function callbacks in general?

[EmmaRenauld]

I'm not sure yet. These are a few methods that I found that were not covered when I was investigating which unit test I should do next. If I find others, I will try to manage them too.

It's only 0.36%, sure, but it still helps me a lot when I want to check if some functions are simply never used anywhere.