bug#40658: Fforward-comment: Incorrect recognition of end of comment in C++

bug#40658: Fforward-comment: Incorrect recognition of end of comment in C++

[Alan Mackenzie]

Hello, Emacs.

C++ comments can be continued onto the next line by an escaped NL.  This
is configured in Emacs with the variable comment-end-can-be-escaped.

However, Fforward_comment (in the backward direction) doesn't take this
into account.  A BOL following an escaped NL cannot be at the end of a
C++ line comment:

    // C++ comment first line \
    Second line of comment
    ^
    |
  point

Point here is in the middle of a comment.

It is therefore wrong in Fforward_comment not to check
comment-end-can-be-escaped when immediately after an Sendcomment.  This
can be fixed thusly:


diff --git a/src/syntax.c b/src/syntax.c
index e24b98da32..ff125b137c 100644
--- a/src/syntax.c
+++ b/src/syntax.c
@@ -2572,8 +2572,9 @@ between them, return t; otherwise return nil.  */)
 	    }
 	  else if (code == Sendcomment)
 	    {
-	      found = back_comment (from, from_byte, stop, comnested, comstyle,
-				    &out_charpos, &out_bytepos);
+              found = (!quoted || !Vcomment_end_can_be_escaped)
+                && back_comment (from, from_byte, stop, comnested, comstyle,
+                                 &out_charpos, &out_bytepos);
 	      if (!found)
 		{
 		  if (c == '\n')


This should clearly go into the master branch.  I think it should also go
into the emacs-27 release branch, since it is relevant for a nasty bug,
bug #40052.  There, scrolling in a C++ include file, which consists of a
#define macro ~2,400 lines long is deadly slow.  With the above fix in
place, scrolling is over 4 times as fast, although still too slow.

-- 
Alan Mackenzie (Nuremberg, Germany).

bug#40658: Fforward-comment: Incorrect recognition of end of comment in C++

[Eli Zaretskii]

> Date: Thu, 16 Apr 2020 12:44:26 +0000
> From: Alan Mackenzie <acm@muc.de>
> Cc: Stefan Monnier <monnier@iro.umontreal.ca>
> 
> This should clearly go into the master branch.  I think it should also go
> into the emacs-27 release branch, since it is relevant for a nasty bug,
> bug #40052.

Sorry, too late for that now.  This code is there for the last 20+
years, we can endure that a bit more, I think.

bug#40658: Fforward-comment: Incorrect recognition of end of comment in C++

[Alan Mackenzie]

Hello, Eli.

On Thu, Apr 16, 2020 at 16:35:14 +0300, Eli Zaretskii wrote:
> > Date: Thu, 16 Apr 2020 12:44:26 +0000
> > From: Alan Mackenzie <acm@muc.de>
> > Cc: Stefan Monnier <monnier@iro.umontreal.ca>

> > This should clearly go into the master branch.  I think it should also go
> > into the emacs-27 release branch, since it is relevant for a nasty bug,
> > bug #40052.

> Sorry, too late for that now.  This code is there for the last 20+
> years, we can endure that a bit more, I think.

OK, I've committed the patch to master.  The code, in effect, just dates
from 2015, when I added support for escaped EOLs in comments.

I will put some workaround into CC Mode (which I would have to have done
anyway).

-- 
Alan Mackenzie (Nuremberg, Germany).

bug#40658: Fforward-comment: Incorrect recognition of end of comment in C++

[Stefan Monnier]

> I will put some workaround into CC Mode

Even better, thanks.

> (which I would have to have done anyway).

Well, there's no such obligation, really ;-)


        Stefan

bug#40658: (Fforward-comment: Incorrect recognition of end of comment in C++)

[Alan Mackenzie]

Bug fixed.

-- 
Alan Mackenzie (Nuremberg, Germany).