Re: [PATCH] Miscellaneous code readability improvements.

[Akim Demaille]

Le 13 août 09 à 21:11, Akim Demaille a écrit :

> > > I think it should be
> > >
> > > > +    complain_at (*loc, _("unrecognized escape sequence: `\\ 
> > > > %s'"), yytext +
> > > > 1);
> > >
> > > to cope with \ followed by a non printable character (including \n).
> >
> > I'm not sure what you're going for.  As far as I can tell, that  
> > just moves
> > the "\" from the argument to the format string.  A following special
> > character like newline still prints as a newline.
>
> Sorry Joel, I meant to restore the call to quote on yytext + 1.

But that's not nice, since in the case of a non printable character,  
say \n, we would display

        `\\n'

which is very misleading.  GCC seems to decide to issue the initial  
backslash when the culprit is printable, otherwise it converts the non- 
printable character into octal and pretends that the octal sequence  
was used, but it does not print the initial backslash:

> const char *cp = "\z\é\ ";

gives:

> foo.cc:2:1: warning: unknown escape sequence '\z'
> foo.cc:2:1: warning: unknown escape sequence: '\303'
> foo.cc:2:1: warning: unknown escape sequence: '\040'

(the inconsistency on the colon looks like an error to me, it's been a  
long time since I last filed a bug report about gcc ;).

Maybe we need to get rid of the backslash from the culprit part:

        invalid character for \-escape: `z'
        invalid character for \-escape: `\n'

?