Re: master 67c3a3a 2/2: Avoid occasional confusion of Flymake C/C++ backend

Re: master 67c3a3a 2/2: Avoid occasional confusion of Flymake C/C++ backend

[Robert Pluim]

>>>>> On Mon, 25 Mar 2019 10:56:34 -0400 (EDT), Unknown <unknown@unknown.invalid> said:

    João> branch: master commit
    João> 67c3a3af1d9e2582193d5ea33b6c190e79ad56e6 Author: João Távora
    João> <joaotavora@gmail.com> Commit: João Távora
    João> <joaotavora@gmail.com>

    João>     Avoid occasional confusion of Flymake C/C++ backend
    
    João>     The regexp would sometimes match non-errors and create a
    João> diagnostic with nil severity, which would have consequences
    João> for the Flymake engine (these should eventually be
    João> addressed, too).

This is making flymake-tests fail:

    Test different-diagnostic-types condition:
        (ert-test-failed
         ((should
           (eq 'flymake-error
                (face-at-point)))
          :form
          (eq flymake-error flymake-note)
          :value nil))
       FAILED  1/9  different-diagnostic-types (1.093461 sec)
       
(face-at-point) is flymake-note there.

Robert

Re: master updated (5fdf4fc -> 67c3a3a)

[Glenn Morris]

Hi, FYI this change causes a test failure on hydra.nixos.
It could be specific to older gcc versions.
Ref: https://hydra.nixos.org/build/91223944

Re: master updated (5fdf4fc -> 67c3a3a)

[Robert Pluim]

>>>>> On Wed, 27 Mar 2019 08:23:34 -0400, Glenn Morris <rgm@gnu.org> said:

    Glenn> Hi, FYI this change causes a test failure on hydra.nixos.
    Glenn> It could be specific to older gcc versions.  Ref:
    Glenn> https://hydra.nixos.org/build/91223944

Does 7.3 count as 'old' these days? Itʼs what's shipped in Ubuntu
18.04.

Robert

Re: master 67c3a3a 2/2: Avoid occasional confusion of Flymake C/C++ backend

[João Távora]

[-- Attachment #1: Type: text/plain, Size: 1210 bytes --]

What is the GCC version that you are using?

João

On Tue, Mar 26, 2019 at 2:01 PM Robert Pluim <rpluim@gmail.com> wrote:

> >>>>> On Mon, 25 Mar 2019 10:56:34 -0400 (EDT), Unknown
> <unknown@unknown.invalid> said:
>
>     João> branch: master commit
>     João> 67c3a3af1d9e2582193d5ea33b6c190e79ad56e6 Author: João Távora
>     João> <joaotavora@gmail.com> Commit: João Távora
>     João> <joaotavora@gmail.com>
>
>     João>     Avoid occasional confusion of Flymake C/C++ backend
>
>     João>     The regexp would sometimes match non-errors and create a
>     João> diagnostic with nil severity, which would have consequences
>     João> for the Flymake engine (these should eventually be
>     João> addressed, too).
>
> This is making flymake-tests fail:
>
>     Test different-diagnostic-types condition:
>         (ert-test-failed
>          ((should
>            (eq 'flymake-error
>                 (face-at-point)))
>           :form
>           (eq flymake-error flymake-note)
>           :value nil))
>        FAILED  1/9  different-diagnostic-types (1.093461 sec)
>
> (face-at-point) is flymake-note there.
>
> Robert
>


-- 
João Távora

[-- Attachment #2: Type: text/html, Size: 1933 bytes --]

Re: master updated (5fdf4fc -> 67c3a3a)

[João Távora]

[-- Attachment #1: Type: text/plain, Size: 1036 bytes --]

Thanks for the heads up, Glenn. I tested with with gcc 8.2

Two questions:

1) can I know exactly which of the two commits caused the
    problem in hydra.nixos.org? (Actually, I think it was 67c3a3 according
to Robert's
    separate message, but in general is it possible on hydra.nixos?)

2) can I have a look at the individual test suite's log files? They
    don't seem to be available at
https://hydra.nixos.org/build/91223944/nixlog/1
    (again I think Robert's email is sufficient diagnostic, but confirming
     this wouldn't hurt).

João

On Wed, Mar 27, 2019 at 12:50 PM Robert Pluim <rpluim@gmail.com> wrote:

> >>>>> On Wed, 27 Mar 2019 08:23:34 -0400, Glenn Morris <rgm@gnu.org> said:
>
>     Glenn> Hi, FYI this change causes a test failure on hydra.nixos.
>     Glenn> It could be specific to older gcc versions.  Ref:
>     Glenn> https://hydra.nixos.org/build/91223944
>
> Does 7.3 count as 'old' these days? Itʼs what's shipped in Ubuntu
> 18.04.
>
> Robert
>
>

-- 
João Távora

[-- Attachment #2: Type: text/html, Size: 2036 bytes --]

Re: master updated (5fdf4fc -> 67c3a3a)

[Dmitry Gutov]

On 27.03.2019 20:04, João Távora wrote:

> Two questions:
> 
> 1) can I know exactly which of the two commits caused the
>      problem in hydra.nixos.org <http://hydra.nixos.org>? (Actually, I 
> think it was 67c3a3 according to Robert's
>      separate message, but in general is it possible on hydra.nixos?)

https://emba.gnu.org/emacs/emacs/pipelines?page=2&scope=all says that it 
was indeed 67c3a3.

Re: master updated (5fdf4fc -> 67c3a3a)

[Robert Pluim]

>>>>> On Wed, 27 Mar 2019 20:23:09 +0200, Dmitry Gutov <dgutov@yandex.ru> said:

    Dmitry> On 27.03.2019 20:04, João Távora wrote:
    >> Two questions:
    >> 
    >> 1) can I know exactly which of the two commits caused the    
    >> problem in hydra.nixos.org <http://hydra.nixos.org>? (Actually,
    >> I think it was 67c3a3 according to Robert's     separate
    >> message, but in general is it possible on hydra.nixos?)

    Dmitry> https://emba.gnu.org/emacs/emacs/pipelines?page=2&scope=all
    Dmitry> says that it was indeed 67c3a3.


As did my git bisect. gcc 7.3, FWIW. Ancient, I know :-)

Robert

Re: master updated (5fdf4fc -> 67c3a3a)

[João Távora]

[-- Attachment #1: Type: text/plain, Size: 1416 bytes --]

On Thu, Mar 28, 2019 at 7:35 AM Robert Pluim <rpluim@gmail.com> wrote:

> >>>>> On Wed, 27 Mar 2019 20:23:09 +0200, Dmitry Gutov <dgutov@yandex.ru>
> said:
>
>     Dmitry> On 27.03.2019 20:04, João Távora wrote:
>     >> Two questions:
>     >>
>     >> 1) can I know exactly which of the two commits caused the
>     >> problem in hydra.nixos.org <http://hydra.nixos.org>? (Actually,
>     >> I think it was 67c3a3 according to Robert's     separate
>     >> message, but in general is it possible on hydra.nixos?)
>
>     Dmitry> https://emba.gnu.org/emacs/emacs/pipelines?page=2&scope=all
>     Dmitry> says that it was indeed 67c3a3.
>

Thanks Dmitry. As I understand there are now two CI systems checking
Emacs but only one of them checks every commit (and that is the emba
one).  Does it also provide access to the `x-tests.log` files?


> As did my git bisect. gcc 7.3, FWIW. Ancient, I know :-)
>

Thanks, it is indeed "worth" something, since problems have arisen due
to different GCC versions before. And it's not a question of being ancient
it's a question of that being valuable information.

Anyway, the tests fail with GCC=8.2 as well: they passed on my local test
run for some reason (probably an oversight with stale .elc on my part).
I'm working on a fix: you can give me a couple of days or revert
67c3a3 if it's becoming too much of a nuisance.

João Távora

[-- Attachment #2: Type: text/html, Size: 2507 bytes --]

Re: master updated (5fdf4fc -> 67c3a3a)

[João Távora]

[-- Attachment #1: Type: text/plain, Size: 2003 bytes --]

On Thu, Mar 28, 2019 at 5:35 PM João Távora <joaotavora@gmail.com> wrote:

> On Thu, Mar 28, 2019 at 7:35 AM Robert Pluim <rpluim@gmail.com> wrote:
>
>> >>>>> On Wed, 27 Mar 2019 20:23:09 +0200, Dmitry Gutov <dgutov@yandex.ru>
>> said:
>>
>>     Dmitry> On 27.03.2019 20:04, João Távora wrote:
>>     >> Two questions:
>>     >>
>>     >> 1) can I know exactly which of the two commits caused the
>>     >> problem in hydra.nixos.org <http://hydra.nixos.org>? (Actually,
>>     >> I think it was 67c3a3 according to Robert's     separate
>>     >> message, but in general is it possible on hydra.nixos?)
>>
>>     Dmitry> https://emba.gnu.org/emacs/emacs/pipelines?page=2&scope=all
>>     Dmitry> says that it was indeed 67c3a3.
>>
>
> Thanks Dmitry. As I understand there are now two CI systems checking
> Emacs but only one of them checks every commit (and that is the emba
> one).  Does it also provide access to the `x-tests.log` files?
>
>
>> As did my git bisect. gcc 7.3, FWIW. Ancient, I know :-)
>>
>
> Thanks, it is indeed "worth" something, since problems have arisen due
> to different GCC versions before. And it's not a question of being ancient
> it's a question of that being valuable information.
>
> Anyway, the tests fail with GCC=8.2 as well: they passed on my local test
> run for some reason (probably an oversight with stale .elc on my part).
> I'm working on a fix: you can give me a couple of days or revert
> 67c3a3 if it's becoming too much of a nuisance.
>

I've fixed the problem.  For those interested, this had to do with failing
to detect the "In file included from xxx.c" errors in the gcc output. Any
number of such conditions for the same file should be posted as an error,
but wasn't.

A better fix would discover individual errors, warnings and notes in an
included .h file and post them all to the .c file but I don't have time for
that
right now.  In case someone does, it shouldn't be very hard to code.

João

[-- Attachment #2: Type: text/html, Size: 3434 bytes --]

Re: master updated (5fdf4fc -> 67c3a3a)

[Ken Brown]

On 3/28/2019 3:05 PM, João Távora wrote:
> On Thu, Mar 28, 2019 at 5:35 PM João Távora <joaotavora@gmail.com 
> <mailto:joaotavora@gmail.com>> wrote:
> 
>     On Thu, Mar 28, 2019 at 7:35 AM Robert Pluim <rpluim@gmail.com
>     <mailto:rpluim@gmail.com>> wrote:
> 
>          >>>>> On Wed, 27 Mar 2019 20:23:09 +0200, Dmitry Gutov
>         <dgutov@yandex.ru <mailto:dgutov@yandex.ru>> said:
> 
>              Dmitry> On 27.03.2019 20:04, João Távora wrote:
>              >> Two questions:
>              >>
>              >> 1) can I know exactly which of the two commits caused the
>              >> problem in hydra.nixos.org <http://hydra.nixos.org>
>         <http://hydra.nixos.org>? (Actually,
>              >> I think it was 67c3a3 according to Robert's     separate
>              >> message, but in general is it possible on hydra.nixos?)
> 
>              Dmitry> https://emba.gnu.org/emacs/emacs/pipelines?page=2&scope=all
>              Dmitry> says that it was indeed 67c3a3.
> 
> 
>     Thanks Dmitry. As I understand there are now two CI systems checking
>     Emacs but only one of them checks every commit (and that is the emba
>     one).  Does it also provide access to the `x-tests.log` files?
> 
>         As did my git bisect. gcc 7.3, FWIW. Ancient, I know :-)
> 
> 
>     Thanks, it is indeed "worth" something, since problems have arisen due
>     to different GCC versions before. And it's not a question of being ancient
>     it's a question of that being valuable information.
> 
>     Anyway, the tests fail with GCC=8.2 as well: they passed on my local test
>     run for some reason (probably an oversight with stale .elc on my part).
>     I'm working on a fix: you can give me a couple of days or revert
>     67c3a3 if it's becoming too much of a nuisance.
> 
> 
> I've fixed the problem.  For those interested, this had to do with failing
> to detect the "In file included from xxx.c" errors in the gcc output. Any
> number of such conditions for the same file should be posted as an error,
> but wasn't.

I can't find where your fix was ever pushed.  Am I missing something?

Ken

Re: master updated (5fdf4fc -> 67c3a3a)

[João Távora]

[-- Attachment #1: Type: text/plain, Size: 2744 bytes --]

On Mon, Jul 1, 2019 at 11:28 PM Ken Brown <kbrown@cornell.edu> wrote:

> On 3/28/2019 3:05 PM, João Távora wrote:
> > On Thu, Mar 28, 2019 at 5:35 PM João Távora <joaotavora@gmail.com
> > <mailto:joaotavora@gmail.com>> wrote:
> >
> >     On Thu, Mar 28, 2019 at 7:35 AM Robert Pluim <rpluim@gmail.com
> >     <mailto:rpluim@gmail.com>> wrote:
> >
> >          >>>>> On Wed, 27 Mar 2019 20:23:09 +0200, Dmitry Gutov
> >         <dgutov@yandex.ru <mailto:dgutov@yandex.ru>> said:
> >
> >              Dmitry> On 27.03.2019 20:04, João Távora wrote:
> >              >> Two questions:
> >              >>
> >              >> 1) can I know exactly which of the two commits caused the
> >              >> problem in hydra.nixos.org <http://hydra.nixos.org>
> >         <http://hydra.nixos.org>? (Actually,
> >              >> I think it was 67c3a3 according to Robert's     separate
> >              >> message, but in general is it possible on hydra.nixos?)
> >
> >              Dmitry>
> https://emba.gnu.org/emacs/emacs/pipelines?page=2&scope=all
> >              Dmitry> says that it was indeed 67c3a3.
> >
> >
> >     Thanks Dmitry. As I understand there are now two CI systems checking
> >     Emacs but only one of them checks every commit (and that is the emba
> >     one).  Does it also provide access to the `x-tests.log` files?
> >
> >         As did my git bisect. gcc 7.3, FWIW. Ancient, I know :-)
> >
> >
> >     Thanks, it is indeed "worth" something, since problems have arisen
> due
> >     to different GCC versions before. And it's not a question of being
> ancient
> >     it's a question of that being valuable information.
> >
> >     Anyway, the tests fail with GCC=8.2 as well: they passed on my local
> test
> >     run for some reason (probably an oversight with stale .elc on my
> part).
> >     I'm working on a fix: you can give me a couple of days or revert
> >     67c3a3 if it's becoming too much of a nuisance.
> >
> >
> > I've fixed the problem.  For those interested, this had to do with
> failing
> > to detect the "In file included from xxx.c" errors in the gcc output. Any
> > number of such conditions for the same file should be posted as an error,
> > but wasn't.
>
> I can't find where your fix was ever pushed.  Am I missing something?
>

Maybe not.  When I looked at my local repo again today, I did find some
perfectly prepared commits from months ago that I have not pushed yet,
and some of them flymake related.  Unfortunately, so much time has
gone by that  I don't know exactly what they are about (and there's no
bug number for this).

Anyway I'll have a better look and probably push them tomorrow.

Sorry about that.
João

[-- Attachment #2: Type: text/html, Size: 4362 bytes --]