bug#20738: 25.0.50; grep/compilation output sometimes has text prop `compilation-message' on final line

bug#20738: 25.0.50; grep/compilation output sometimes has text prop `compilation-message' on final line

[Drew Adams]

I seem not to see the problem if there are not a lot of lines output by
`grep', but if there are a lot then I see the problem (don't ask me how
much is a lot - I certainly see it if there are, say, 6000 lines).

If I test `M-: (get-text-property (point) 'compilation-message)' with
point at bol of the last line of the buffer, which is for example, this:

Grep finished (matches found) at Thu Jun  4 23:26:06

Then I get this, for example:

[cl-struct-compilation--message
 (nil 26
      (("Grep finished (matches found) at Thu Jun  4 23" nil)
       nil
       (26 #1))
      nil nil)
 2 nil]

IIUC, there should be no such text property on this line, as it is not
an "error" message line.

Here is what `C-u C-x =' shows wrt text properties at that position:

There are text properties here:
  compilation-handle-exit t
  compilation-message  [Show]
  face                 nil
  font-lock-face       (compilation-info)
  fontified            t
  help-echo            "mouse-2: visit this file and line"
  keymap               compilation-button-map
  mouse-face           compilation-mouseover

And clicking [Show] shows the vector value I wrote above.


In GNU Emacs 25.0.50.1 (i686-pc-mingw32)
 of 2014-10-20 on LEG570
Bzr revision: 118168 rgm@gnu.org-20141020195941-icp42t8ttcnud09g
Windowing system distributor `Microsoft Corp.', version 6.1.7601
Configured using:
 `configure --enable-checking=yes,glyphs CPPFLAGS=-DGLYPH_DEBUG=1'

bug#20738: 25.0.50; grep/compilation output sometimes has text prop `compilation-message' on final line

[Drew Adams]

I said:

> I seem not to see the problem if there are not a lot of lines output
> by `grep', but if there are a lot then I see the problem (don't ask
> me how much is a lot - I certainly see it if there are, say, 6000 lines).

Actually, it's not necessarily the case that the problem occurs only
when there are a lot of output lines.  It happens sometimes and not
other times, but it's not clear to me what the criteria are.

Also, I can reproduce it with Emacs versions as far back as 24.1
(I haven't tried further back.  I think 24.1 is maybe when property
`compilation-message' was first introduced anyway.)

bug#20738: 25.0.50; grep/compilation output sometimes has text prop `compilation-message' on final line

[Lars Ingebrigtsen]

Drew Adams <drew.adams@oracle.com> writes:

> I said:
>
>> I seem not to see the problem if there are not a lot of lines output
>> by `grep', but if there are a lot then I see the problem (don't ask
>> me how much is a lot - I certainly see it if there are, say, 6000 lines).
>
> Actually, it's not necessarily the case that the problem occurs only
> when there are a lot of output lines.  It happens sometimes and not
> other times, but it's not clear to me what the criteria are.

I've tried various greps (in the lisp directory), grepping for "defun"
or "def", the latter giving 35K hits, and I'm unable to reproduce this
bug.

Do you have a recipe?

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

bug#20738: 25.0.50; grep/compilation output sometimes has text prop `compilation-message' on final line

[Drew Adams]

> Do you have a recipe?

No.

bug#20738: 25.0.50; grep/compilation output sometimes has text prop `compilation-message' on final line

[Lars Ingebrigtsen]

Drew Adams <drew.adams@oracle.com> writes:

>> Do you have a recipe?
>
> No.

Then it seems unlikely that we'll make further progress here, and I'm
closing this bug report.

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

bug#20738: 25.0.50; grep/compilation output sometimes has text prop `compilation-message' on final line

[Drew Adams]

> >> Do you have a recipe?
> >
> > No.
> 
> Then it seems unlikely that we'll make further progress here, and I'm
> closing this bug report.

Dunno how you figure that it's unlikely.  I said,
at the outset:

  It happens sometimes and not other times, but it's not
  clear to me what the criteria are.

  Also, I can reproduce it with Emacs versions as far
  back as 24.1  (I haven't tried further back.  I think
  24.1 is maybe when property `compilation-message' was
  first introduced anyway.)

The bug exists.  I don't have recipe to repro it.

There's no relation between not having a recipe and the
frequency of the problem occurring - zero.

The choices, IMO, are to either (1) leave the bug open,
hoping that at some point someone will find or stumble
upon a recipe or a solution or (2) close it, knowing
that the bug exists.

It's not uncommon for bugs to later get more info and
be solved (#1).  Of course, if the real aim is to keep
down the bug count, as opposed to recording bugs in
hopes of fixing them, then #2 can happen.

That sometimes happens in organizations where QA is
judged by the outstanding bug count.  (IMHO, that has
no place in free software.)