bug#35254: 27.0.50; cc-mode/electric-pair-mode/electric-layout-mode: bad trailing whitespace behavior in cc-mode

["João Távora" <joaotavora@gmail.com>]

[-- Attachment #1: Type: text/plain, Size: 2237 bytes --]

On Tue, May 14, 2019 at 12:38 AM Noam Postavsky <npostavs@gmail.com> wrote:

>
> > 1. Does it fix the reported problem (assuming it is a problem, and not
> > an otherwise potentially desirable change in behaviour)?
>
> It does fix the problem.
>

It reintroduces the previous behaviour, I gather. Can you explain
quickly why it was "a problem"?


> > 2. Do any of you have suspicions that it might introduce problems
> > elsewhere?
>
> I'm unsure.  It seems to be undoing a small part of [fd94312443]
> 2019-01-22 "electric-layout-mode kicks in before electric-pair-mode", so
> I guess it might rebreak whatever that commit is fixing.  But I don't
> quite understand what that commit is fixing (in particular, where the
> commit message says "which can be a problem in some modes", which modes
> are those?  What is "a problem"?).
>

Sorry, can't say without investigating much more than time allows. Can you
post the complete sentence?

I vaguely remember that if electric-pair-mode kicked in before
electric-layout-mode we would need more complex layout specs and more
painful indentation logic.  That's why I changed it.  There is a thread of
discussion with Stefan somewhere about this, not sure if public or off-list.


> > 3. Does it pass the automated test suite?
>
> No, it breaks 3 tests in tests/lisp/electric.el:
>
> 3 unexpected results:
>    FAILED  electric-layout-int-main-kernel-style
>    FAILED  electric-layout-plainer-c-mode-use-c-style
>    FAILED  electric-modes-int-main-allman-style
>
> In each case, the reason for failure is that the expected result has
> trailing whitespace that the actual result misses.  I guess
> electric-layout does want to put trailing whitespace in certain cases?
>

Yes, it certainly does.  That trailing whitespace is indentation, I
believe. And
the cursor should be left at that indentation.  Can you confirm? Anyway, if
it's
breaking tests it's almost certainly not what we want.  And if it breaks in
"plainer-c-mode" (a slightly better behaved c-mode), then its even more
certain that it's not what we want.

... unless the tests are demading something unreasonable from the electric
modes, of course.

-- 
João Távora

[-- Attachment #2: Type: text/html, Size: 3239 bytes --]