Re: [dev] dead acute not working on release 0.8.3

From: Christian Tenllado <ctenllado_AT_gmail.com>
Date: Tue, 12 May 2020 19:03:25 +0200

Thank you Jakub for pointing me out the git bisect tool, I hadn't used
it before, very useful.

I have searched for the commit responsible of the bug but there are some
commits for which st does not compile. I did not try to fix the compile
errors, but did some trials assuming good/bad for those commits to
narrow down the problem. I copied the results of the bisect trials
below, and the conclusions of my search at the end. I narrowed down the
problem to 3 commits for which I could not compile st,
that are in between the last confirmed good and the first confirmed bad commits.
I hope this helps.

Regards,

Christian

First bisect trial:
7ceb3d1 -> good
5703aa0 -> bad
99de333 -> does not compile, assume bad
2e54a21 -> good
895e5b5 -> good -> 99de33395126fc9708f442d155e737b9182f6ef4

Second bisect trial:
7ceb3d1 -> good
5703aa0 -> bad
99de333 -> does not compile, assume good
26cdfeb -> bad
cd78575 -> does not compile, assume good -> 26cdfebf31f024e331429e482b1ee342708888e3

Third bisect trial:
7ceb3d1 -> good
5703aa0 -> bad
99de333 -> does not compile, assume good
26cdfeb -> bad
cd78575 -> does not compile, assume bad
2cb5391 -> does not compile, assume good -> cd785755f2e3e3305c7d2556a04423a40bce060a

Fourth bisect trial:
7ceb3d1 -> good
5703aa0 -> bad
99de333 -> does not compile, assume good
26cdfeb -> bad
cd78575 -> does not compile, assume bad
2cb5391 -> does not compile, assume bad -> 2cb539142b97bd2a5c1a322fd7c063c6afb67c9b

CONCLUSIONS:
Most recent bad confirmed: 26cdfeb
Most recent good confirmed: 895e5b5
3 commits in between, that do not compile

git log --oneline 895e5b5~..26cdfeb
26cdfeb (HEAD) x: fix XIM handling
cd78575 x: check we still have an XIC context before accessing it
2cb5391 x: do not instantiate a new nested list on each cursor move
99de333 x: move IME variables into XWindow ime embedded struct
895e5b5 Increase XmbLookupString buffer

Excerpts from Jakub Leszczak's message of May 12, 2020 5:10 pm:
>> Can I provide any extra debug info?
>
> You could. You could do a bisection to find out which exact commit
> introduced regression.
> There is ~50 commits between 0.8.2 and 0.8.3, so you would need to
> compile and test st ~6 times to find out exact commit.
>
> You can use `git bisect` or eyeball more or less middle of `git log
> --oneline`, then `git checkout COMMIT`, compile and test.
> If COMMIT works it means bug is in later commit, and if it doesn't
> work it means that bug is in this or some previous commit.
> Repeat couple times and voila. This could be very helpful.
>
> Br,
> Jakub Leszczak
>
>
Received on Tue May 12 2020 - 19:03:25 CEST