Re: [Emacs-diffs] emacs-25 326ffcc: Allow line comments ending with escaped NL to be continued to the next line.

Re: [Emacs-diffs] emacs-25 326ffcc: Allow line comments ending with escaped NL to be continued to the next line.

[Stefan Monnier]

> -      /* Ignore escaped characters, except comment-enders.  */
> -      if (code != Sendcomment && char_quoted (from, from_byte))
> +      /* Ignore escaped characters, except comment-enders which cannot
> +         be escaped.  */
> +      if ((Vcomment_end_can_be_escaped || code != Sendcomment)
> +          && char_quoted (from, from_byte))
[...]
>        if (code == Sendcomment
>  	  && SYNTAX_FLAGS_COMMENT_STYLE (syntax, 0) == style
>  	  && (SYNTAX_FLAGS_COMMENT_NESTED (syntax) ?
> -	      (nesting > 0 && --nesting == 0) : nesting < 0))
> +	      (nesting > 0 && --nesting == 0) : nesting < 0)
> +          && !(Vcomment_end_can_be_escaped && char_quoted (from, from_byte)))

Hmm... the way I read this, it would mean that

        /*blabla\*/

will be treated as an unclosed comment.  Is that the way C defines it as well?


        Stefan

Re: [Emacs-diffs] emacs-25 326ffcc: Allow line comments ending with escaped NL to be continued to the next line.

[Paul Eggert]

Stefan Monnier wrote:
> Hmm... the way I read this, it would mean that
>
>          /*blabla\*/
>
> will be treated as an unclosed comment.  Is that the way C defines it as well?

No, that's a closed comment. More interestingly, this:

/\
\
*blabla*\
\
/
printf ("\
\\
n\
");

is a closed comment followed by printf ("\n"), because backslash-newline removal 
is done before tokenization. So the "right" fix would be to ignore 
backslash-newline before identifying comments and other tokens.

Backslash-backslash-newline can be tricky, as I recall from long ago when I was 
hacking on the GNU C preprocessor.

Re: [Emacs-diffs] emacs-25 326ffcc: Allow line comments ending with escaped NL to be continued to the next line.

[Alan Mackenzie]

Hello, Stefan.

On Mon, Dec 28, 2015 at 11:14:36AM -0500, Stefan Monnier wrote:
> > -      /* Ignore escaped characters, except comment-enders.  */
> > -      if (code != Sendcomment && char_quoted (from, from_byte))
> > +      /* Ignore escaped characters, except comment-enders which cannot
> > +         be escaped.  */
> > +      if ((Vcomment_end_can_be_escaped || code != Sendcomment)
> > +          && char_quoted (from, from_byte))
> [...]
> >        if (code == Sendcomment
> >  	  && SYNTAX_FLAGS_COMMENT_STYLE (syntax, 0) == style
> >  	  && (SYNTAX_FLAGS_COMMENT_NESTED (syntax) ?
> > -	      (nesting > 0 && --nesting == 0) : nesting < 0))
> > +	      (nesting > 0 && --nesting == 0) : nesting < 0)
> > +          && !(Vcomment_end_can_be_escaped && char_quoted (from, from_byte)))

> Hmm... the way I read this, it would mean that

>         /*blabla\*/
                   ^
		   |
> will be treated as an unclosed comment.  Is that the way C defines it as well?

I've tried it, it's not the case.  The indicated * doesn't have
Sendcomment syntax, so the if statement doesn't trigger.

>         Stefan

-- 
Alan Mackenzie (Nuremberg, Germany).

Re: [Emacs-diffs] emacs-25 326ffcc: Allow line comments ending with escaped NL to be continued to the next line.

[Alan Mackenzie]

Hello, Paul.

On Mon, Dec 28, 2015 at 08:41:23AM -0800, Paul Eggert wrote:
> Stefan Monnier wrote:
> > Hmm... the way I read this, it would mean that
> >
> >          /*blabla\*/
> >
> > will be treated as an unclosed comment.  Is that the way C defines it as well?

> No, that's a closed comment. More interestingly, this:

> /\
> \
> *blabla*\
> \
> /
> printf ("\
> \\
> n\
> ");

> is a closed comment followed by printf ("\n"), because backslash-newline removal 
> is done before tokenization. So the "right" fix would be to ignore 
> backslash-newline before identifying comments and other tokens.

In general, that's probably more hassle than it's worth.  But I don't
think it would be too difficult to remove escaped newlines from
comments.  I'll see what I can manage.

> Backslash-backslash-newline can be tricky, as I recall from long ago when I was 
> hacking on the GNU C preprocessor.

Yes.

-- 
Alan Mackenzie (Nuremberg, Germany).