Re: master b811ad1: Improve behavior of 'electric-pair-mode' in 'cc-mode'

Re: master b811ad1: Improve behavior of 'electric-pair-mode' in 'cc-mode'

[Lars Ingebrigtsen]

larsi@gnus.org (Lars Ingebrigtsen) writes:

> commit b811ad15beeb4b96acb9d83b3a5770ef9f5fe172
> Author: Jim Porter <jporterbugs@gmail.com>
> Commit: Lars Ingebrigtsen <larsi@gnus.org>

I'm seeing some intermittent test failures after this change (I think),
but it's frustratingly difficult to pin down.  It only seems to happen
when I say "make check-maybe", but not otherwise.

Is anybody else seeing this?

-- 
(domestic pets only, the antidote for overdose, milk.)
   bloggy blog: http://lars.ingebrigtsen.no

Re: master b811ad1: Improve behavior of 'electric-pair-mode' in 'cc-mode'

[Jim Porter]

On 11/5/2021 9:21 PM, Lars Ingebrigtsen wrote:
> larsi@gnus.org (Lars Ingebrigtsen) writes:
> 
>> commit b811ad15beeb4b96acb9d83b3a5770ef9f5fe172
>> Author: Jim Porter <jporterbugs@gmail.com>
>> Commit: Lars Ingebrigtsen <larsi@gnus.org>
> 
> I'm seeing some intermittent test failures after this change (I think),
> but it's frustratingly difficult to pin down.  It only seems to happen
> when I say "make check-maybe", but not otherwise.
> 
> Is anybody else seeing this?

I see this too, but as you say, it only happens with "make check-maybe" 
(or when running electric-tests.el manually *after* "make check-maybe"). 
I can't reproduce it otherwise. If I check out the commit before this 
patch, the issue goes away though.

I'm not totally clear on what check-maybe does, but it seems to create 
some persistent state that causes the electric-tests to fail. The 
failure I see is in `whitespace-chomping-dont-cross-comments' for 
`c-mode', which would seem to point to my patch being the culprit, but 
if I'm understanding what that function is testing, everything works as 
it should when I try it out interactively.

- Jim

Re: master b811ad1: Improve behavior of 'electric-pair-mode' in 'cc-mode'

[Lars Ingebrigtsen]

Jim Porter <jporterbugs@gmail.com> writes:

> I see this too, but as you say, it only happens with "make
> check-maybe" (or when running electric-tests.el manually *after* "make
> check-maybe"). I can't reproduce it otherwise. If I check out the
> commit before this patch, the issue goes away though.

I see it if some parts of the Emacs source changes -- then check-maybe
reports a failure.  But it goes away after a full check.  But doesn't
come back if I say check-maybe and electric-tests.  :-/

> I'm not totally clear on what check-maybe does, but it seems to create
> some persistent state that causes the electric-tests to fail.

Yes, it must be doing that, but I'm also vague on what it actually does.
Is there anybody with any insights here?

> The failure I see is in `whitespace-chomping-dont-cross-comments' for
> `c-mode', which would seem to point to my patch being the culprit, but
> if I'm understanding what that function is testing, everything works
> as it should when I try it out interactively.

Yes, I'm not seeing any issues either when testing interactively.

The electric tests are quite involved -- I've added João to the CCs.
Perhaps there's something extra that has to be done here for those.

-- 
(domestic pets only, the antidote for overdose, milk.)
   bloggy blog: http://lars.ingebrigtsen.no