multimail/p4: issues identified by lgtm.com

Johannes Schindelin· Mar 19, 2018, 11:24 UTC · lore

Hi team,

while Coverity was down (it still is not reachable for me, but I guess that's just because everybody and their dog wants to catch up on a month of work delayed by their outage), I tried to find alternatives, and one of them is lgtm.com. Their C/C++ support is not generally available yet, but there have been a couple of issues identified in our Python scripts:

https://lgtm.com/projects/g/git/git/alerts/?mode=list&rule=py%2Fcall%2Fwrong-arguments%2Cpy%2Fcall%2Fwrong-named-argument%2Cpy%2Fexplicit-call-to-delete%2Cpy%2Finconsistent-equality%2Cpy%2Finheritance%2Fincorrect-overridden-signature%2Cpy%2Fmissing-call-to-init%2Cpy%2Fmultiple-definition%2Cpy%2Fredundant-else%2Cpy%2Funinitialized-local-variable%2Cpy%2Funreachable-statement

From a cursory look, it *does* seem as if there are legitimate issues there that want to be fixed ;-) But then, I am far from a Python expert, so my impressions could be completely off the mark.

Ciao, Johannes

Matthieu Moy· Mar 19, 2018, 13:50 UTC · re: Johannes Schindelin · lore

Re: multimail/p4: issues identified by lgtm.com

Johannes Schindelin writes:

Show 13 quoted lines

> Hi team,
>
> while Coverity was down (it still is not reachable for me, but I guess
> that's just because everybody and their dog wants to catch up on a month
> of work delayed by their outage), I tried to find alternatives, and one of
> them is lgtm.com. Their C/C++ support is not generally available yet, but
> there have been a couple of issues identified in our Python scripts:
>
> https://lgtm.com/projects/g/git/git/alerts/?mode=list&rule=py%2Fcall%2Fwrong-arguments%2Cpy%2Fcall%2Fwrong-named-argument%2Cpy%2Fexplicit-call-to-delete%2Cpy%2Finconsistent-equality%2Cpy%2Finheritance%2Fincorrect-overridden-signature%2Cpy%2Fmissing-call-to-init%2Cpy%2Fmultiple-definition%2Cpy%2Fredundant-else%2Cpy%2Funinitialized-local-variable%2Cpy%2Funreachable-statement
>
> From a cursory look, it *does* seem as if there are legitimate issues
> there that want to be fixed ;-) But then, I am far from a Python expert,
> so my impressions could be completely off the mark.

For git-multimail, there are actually a lot of false-positive due to the way we use multiple-inheritance (when calling super(...).__init__(), lgtm assumes we're calling the superclass's constructor while we actually call the "next in the method resolution order list"), but also probably valid concerns.

I've set up a separate project for git-multimail:

  https://lgtm.com/projects/g/git-multimail/git-multimail/alerts/?mode=list

I'll have a deeper look ASAP.

Thanks!

-- 
Matthieu Moy
https://matthieu-moy.fr/