Fixing midx bitmap related racy test failure (under sha-256)

From: Abhradeep Chakraborty <chakrabortyabhradeep79@gmail.com>

Hello developers,

I have been writing a patch series[1] to integrate a lookup table extension to .bitmap file. I reformatted the test code so that I can run the same tests for the lookup table extension. But this is causing a failure in a midx bitmap related test case[2]. It is not related to my code because this can be produced with out my changes.

If you see the test file[3] (with my changes) and also observe the test results[4], you'll notice that the said test case is always failing on the second `test_midx_bitmap_cases` call under sha-256 only. This is racy in nature. I.e. it might pass after executing the test 2 or 3 times.

I have been trying to solve the issue for one or two week but as I am not very clear about the internal working of multi-pack-index or repacking, I am still not able to solve the issue. Besides, my semester exam is starting from 18 Aug. So I have to give a significant amount of time to studies.

So, it will be very helpful if anyone help me to find out the solution. You can know more from this thread[5].

Thanks :)

[1] https://lore.kernel.org/git/pull.1266.git.1655728395.gitgitgadget@gmail.com/ [2] https://github.com/Abhra303/git/blob/b460516b306e6885cd1c0af1c3379fb953952de2/t/lib-bitmap.sh#L195 [3] https://github.com/Abhra303/git/blob/bitmap-commit-table/t/t5326-multi-pack-bitmaps.sh [4] https://github.com/Abhra303/git/runs/7827604510?check_suite_focus=true [5] https://lore.kernel.org/git/p3r70610-8n52-s8q0-n641-onp4ps01330n@tzk.qr/#t