"ignored" in *vc-dir* buffers

"ignored" in *vc-dir* buffers

[Lars Ingebrigtsen]

When I'm checking stuff in in Emacs 28 now, I'm always getting

                         ./
                         doc/misc/
  *  up-to-date          doc/misc/eww.texi
                         etc/
  *  up-to-date          etc/NEWS
                         lisp/net/
  *  ignored             lisp/net/shr.el

or something like that.  That is, a few of the lines will say "ignored"
while the rest say "up-to-date" as expected.

As far as I can tell, nothing is actually ignored...

This used to work better before...  is this something other people are
seeing, or should I start debugging?

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

Re: "ignored" in *vc-dir* buffers

[Dmitry Gutov]

On 17.07.2020 16:48, Lars Ingebrigtsen wrote:
> When I'm checking stuff in in Emacs 28 now, I'm always getting
> 
>                           ./
>                           doc/misc/
>    *  up-to-date          doc/misc/eww.texi
>                           etc/
>    *  up-to-date          etc/NEWS
>                           lisp/net/
>    *  ignored             lisp/net/shr.el
> 
> or something like that.  That is, a few of the lines will say "ignored"
> while the rest say "up-to-date" as expected.
> 
> As far as I can tell, nothing is actually ignored...
> 
> This used to work better before...  is this something other people are
> seeing, or should I start debugging?

Can't reproduce. If I edit shr.el and save it, it shows up as "edited" 
in the VC-Dir buffer.

Re: "ignored" in *vc-dir* buffers

[Lars Ingebrigtsen]

Dmitry Gutov <dgutov@yandex.ru> writes:

> On 17.07.2020 16:48, Lars Ingebrigtsen wrote:
>> When I'm checking stuff in in Emacs 28 now, I'm always getting
>>                           ./
>>                           doc/misc/
>>    *  up-to-date          doc/misc/eww.texi
>>                           etc/
>>    *  up-to-date          etc/NEWS
>>                           lisp/net/
>>    *  ignored             lisp/net/shr.el
>> or something like that.  That is, a few of the lines will say
>> "ignored"
>> while the rest say "up-to-date" as expected.
>> As far as I can tell, nothing is actually ignored...
>> This used to work better before...  is this something other people
>> are
>> seeing, or should I start debugging?
>
> Can't reproduce. If I edit shr.el and save it, it shows up as "edited"
> in the VC-Dir buffer.

Yes, that works correctly.  It's only when I check files in that they
switch to "ignored".

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

Re: "ignored" in *vc-dir* buffers

[Dmitry Gutov]

On 17.07.2020 17:25, Lars Ingebrigtsen wrote:
> Yes, that works correctly.  It's only when I check files in that they
> switch to "ignored".

Always?

I added a change to shr.el, made a commit. It because 'up-to-date'.

Re: "ignored" in *vc-dir* buffers

[Lars Ingebrigtsen]

Dmitry Gutov <dgutov@yandex.ru> writes:

> On 17.07.2020 17:25, Lars Ingebrigtsen wrote:
>> Yes, that works correctly.  It's only when I check files in that they
>> switch to "ignored".
>
> Always?
>
> I added a change to shr.el, made a commit. It because 'up-to-date'.

Only some of the files...

OK, I'll try to debug and see what's up.

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

Re: "ignored" in *vc-dir* buffers

[Dmitry Gutov]

Sorry,

On 17.07.2020 18:17, Dmitry Gutov wrote:
> On 17.07.2020 17:25, Lars Ingebrigtsen wrote:
>> Yes, that works correctly.  It's only when I check files in that they
>> switch to "ignored".
> 
> Always?
> 
> I added a change to shr.el, made a commit. It because 'up-to-date'.

                                                 ^ became

Re: "ignored" in *vc-dir* buffers

[Lars Ingebrigtsen]

OK, I've now debugged a bit.

If I edit a file under the lisp directory (but only there), I get these
weird results.

First, if I edit a file, I get the expected:

(vc-git-state "/home/larsi/src/emacs/trunk/lisp/vc/vc-git.el")
=> edited

But then if I remove the edit, then:

(vc-git-state "/home/larsi/src/emacs/trunk/lisp/vc/vc-git.el")
=> ignored

Outside the lisp directory, I get the expected:

(vc-git-state "/home/larsi/src/emacs/trunk/src/gnutls.c")
=> up-to-date

This is with "git version 2.27.0" in Debian sid.

And, indeed:

larsi@xo:~/src/emacs/trunk$ git status --porcelain --ignored -- lisp/vc/vc-git.el
!! lisp/vc/vc-git.elc

So...  er...  it's not an Emacs issue at all, but my git is broken?
Because "!!" means ignored.  Or is my Emacs tree broken?

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

Re: "ignored" in *vc-dir* buffers

[Stefan Monnier]

> larsi@xo:~/src/emacs/trunk$ git status --porcelain --ignored -- lisp/vc/vc-git.el
> !! lisp/vc/vc-git.elc
                     ^^^
What's this "c"?


        Stef

Re: "ignored" in *vc-dir* buffers

[Lars Ingebrigtsen]

Stefan Monnier <monnier@iro.umontreal.ca> writes:

>> larsi@xo:~/src/emacs/trunk$ git status --porcelain --ignored --
>> lisp/vc/vc-git.el
>> !! lisp/vc/vc-git.elc
>                      ^^^
> What's this "c"?

Oh yeah!  Good catch.

It looks like this git is interpreting the path spec not as an absolute
name, but as a prefix?

larsi@xo:~/src/emacs/trunk$ git status --porcelain --untracked-files --ignored -- src/gnutls
!! src/gnutls.o

Yup!

larsi@xo:~/src/emacs/trunk$ git status --porcelain --untracked-files --ignored -- src/g
!! src/gl-stamp
!! src/globals.h
!! src/globals.h.~1~
!! src/gnutls.o
!! src/gtkutil.o

Older versions of git doesn't interpret PATHSPEC this way, and I can't
see that the manual says anything about this, either...

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

Re: "ignored" in *vc-dir* buffers

[Andreas Schwab]

On Jul 19 2020, Lars Ingebrigtsen wrote:

> Older versions of git doesn't interpret PATHSPEC this way, and I can't
> see that the manual says anything about this, either...

It's a recent (2.27) regression.

https://lore.kernel.org/git/87lfjg6mkn.fsf@linux-m68k.org/

Andreas.

-- 
Andreas Schwab, schwab@linux-m68k.org
GPG Key fingerprint = 7578 EB47 D4E5 4D69 2510  2552 DF73 E780 A9DA AEC1
"And now for something completely different."