bug#28428: comment-search-backward with no comments

bug#28428: comment-search-backward with no comments

[N. Raghavendra]

[-- Attachment #1: Type: text/plain, Size: 1502 bytes --]

I am using Emacs 26.0.50.  In the cases that I tried, when there is no
comment in the portion of the buffer between the search limit and point,
`comment-search-backward' does not move point to the search limit before
raising an error.  This behaviour contradicts its docstring.  My message
to emacs-devel about this is at
http://lists.gnu.org/archive/html/emacs-devel/2017-09/msg00193.html

Attached are a patch that seems to fix the problem, and a file
containing a test.  Running the test without the change in the patch
fails, and passes with the change:

$ git checkout master

$ emacs -Q -batch -l lisp/newcomment.el -l /tmp/test.el \
  -f ert-run-tests-batch-and-exit
----------------------------------------------------------------------
(ert-test-failed
     ((should
       (=
        (point)
        limit))
      :form
      (= 70 41)
      :value nil))
   FAILED  1/1  test-comment-search-backward-absent
----------------------------------------------------------------------

$ git checkout test-comment-search-forward-absent

$ emacs -Q -batch -l lisp/newcomment.el -l /tmp/test.el \
  -f ert-run-tests-batch-and-exit
----------------------------------------------------------------------
Running 1 tests (2017-09-12 03:31:48+0530)
   passed  1/1  test-comment-search-backward-absent
----------------------------------------------------------------------

Raghu.

--
N. Raghavendra <raghu@hri.res.in>, http://www.retrotexts.net/
Harish-Chandra Research Institute, http://www.hri.res.in/


[-- Warning: decoded text below may be mangled, UTF-8 assumed --]
[-- Attachment #2: 0001-comment-search-backword-move-point-when-there-is-no-comment.patch --]
[-- Type: text/x-diff, Size: 1056 bytes --]

From eb8e32c6fe4a3d08be8b856b3ff4a300a2e69d4b Mon Sep 17 00:00:00 2001
From: "N. Raghavendra" <raghu@hri.res.in>
Date: Tue, 12 Sep 2017 01:41:16 +0530
Subject: [PATCH] comment-search-backword: move point when there is no comment

* lisp/newcomment.el (comment-search-backward): Move point to
LIMIT when there is no comment between LIMIT and point.
---
 lisp/newcomment.el | 4 +++-
 1 file changed, 3 insertions(+), 1 deletion(-)

diff --git a/lisp/newcomment.el b/lisp/newcomment.el
index 8772b523..d849fc1 100644
--- a/lisp/newcomment.el
+++ b/lisp/newcomment.el
@@ -525,7 +525,9 @@ comment-search-backward
   ;; comment-set-column) and to find the comment-start string (via
   ;; comment-beginning) in indent-new-comment-line, it should be harmless.
   (if (not (re-search-backward comment-start-skip limit t))
-      (unless noerror (error "No comment"))
+      (progn
+        (goto-char limit)
+        (unless noerror (error "No comment")))
     (beginning-of-line)
     (let* ((end (match-end 0))
 	   (cs (comment-search-forward end t))
-- 
2.7.4


[-- Attachment #3: test.el --]
[-- Type: application/emacs-lisp, Size: 667 bytes --]

bug#28428: comment-search-backward with no comments

[Stefan Monnier]

>    (if (not (re-search-backward comment-start-skip limit t))
> -      (unless noerror (error "No comment"))
> +      (progn
> +        (goto-char limit)
> +        (unless noerror (error "No comment")))

That looks fine, thank you.  But I think the simpler patch below works
as well:

    diff --git a/lisp/newcomment.el b/lisp/newcomment.el
    index be6dbe3a4c..e32966e596 100644
    --- a/lisp/newcomment.el
    +++ b/lisp/newcomment.el
    @@ -524,7 +524,7 @@ comment-search-backward
       ;; comment-search-backward is only used to find the comment-column (in
       ;; comment-set-column) and to find the comment-start string (via
       ;; comment-beginning) in indent-new-comment-line, it should be harmless.
    -  (if (not (re-search-backward comment-start-skip limit t))
    +  (if (not (re-search-backward comment-start-skip limit 'move))
           (unless noerror (error "No comment"))
         (beginning-of-line)
         (let* ((end (match-end 0))

I wonder if there's code out there that depends on this behavior, tho,
since AFAIK it's been behaving this way "forever".


        Stefan

bug#28428: comment-search-backward with no comments

[N. Raghavendra]

At 2017-09-22T12:40:58-04:00, Stefan Monnier wrote:

> That looks fine, thank you.  But I think the simpler patch below works
> as well:
>
>     -  (if (not (re-search-backward comment-start-skip limit t))
>     +  (if (not (re-search-backward comment-start-skip limit 'move))

That's indeed nicer!  Thanks for pointing out that the NOERROR arg of
`re-search-backward' can be given a value outside {t,nil} to move point
to LIMIT.  It's a lesson to myself that I should read docstrings more
carefully.

> I wonder if there's code out there that depends on this behavior, tho,
> since AFAIK it's been behaving this way "forever".

As you say, it doesn't really matter; it's just that there is a
discrepancy between the docstring specification of the effects of the
function and the said effects, which would be good to remove.

I came across it because I've been playing around with a major mode
based on AUCTeX, and was writing ERT tests for some initialisation code
I was using from AUCTeX.  Perhaps such tests are the only places where
these trivial errors matter.

Regards,
Raghu.

-- 
N. Raghavendra <raghu@hri.res.in>, http://www.retrotexts.net/
Harish-Chandra Research Institute, http://www.hri.res.in/

bug#28428: comment-search-backward with no comments

[Stefan Monnier]

>> That looks fine, thank you.  But I think the simpler patch below works
>> as well:
>> -  (if (not (re-search-backward comment-start-skip limit t))
>> +  (if (not (re-search-backward comment-start-skip limit 'move))
> That's indeed nicer!  Thanks for pointing out that the NOERROR arg of
> `re-search-backward' can be given a value outside {t,nil} to move point
> to LIMIT.  It's a lesson to myself that I should read docstrings more
> carefully.

I think it's too risky for the emacs-26 branch, but feel free to install
it on master,


        Stefan

bug#28428: comment-search-backward with no comments

[N. Raghavendra]

At 2017-09-22T15:26:58-04:00, Stefan Monnier wrote:

> I think it's too risky for the emacs-26 branch, but feel free to install
> it on master,

Do you mean that the change can be pushed to master?  If so, do you want
me to do anything about it?  I guess someone with commit access will
have to take over.

Raghu.

-- 
N. Raghavendra <raghu@hri.res.in>, http://www.retrotexts.net/
Harish-Chandra Research Institute, http://www.hri.res.in/

bug#28428: comment-search-backward with no comments

[Stefan Monnier]

Version: 27.1

I pushed the patch to `master`.


        Stefan

bug#28428: comment-search-backward with no comments

[N. Raghavendra]

At 2017-09-23T17:07:15-04:00, Stefan Monnier wrote:

> I pushed the patch to `master`.

Perfect!

Thanks,
Raghu.

-- 
N. Raghavendra <raghu@hri.res.in>, http://www.retrotexts.net/
Harish-Chandra Research Institute, http://www.hri.res.in/