[MPlayer-dev-eng] [PATCH] fix compile with clang r150242

[Diego Biurrun]

On Mon, Feb 27, 2012 at 11:46:57PM +0200, Ivan Kalvachev wrote:
> On 2/27/12, Diego Biurrun <diego at biurrun.de> wrote:
> > On Mon, Feb 27, 2012 at 09:58:02PM +0200, Ivan Kalvachev wrote:
> >> On 2/27/12, Diego Biurrun <diego at biurrun.de> wrote:
> >> > On Sat, Feb 25, 2012 at 11:48:45PM +0200, Ivan Kalvachev wrote:
> >> >> On 2/25/12, Nicolas George <nicolas.george at normalesup.org> wrote:
> >> >> > Le sextidi 6 ventôse, an CCXX, compn a écrit :
> >> >> >> sub/sub_cc.c:68:19: error: illegal character encoding in character
> >> >> >> literal
> >> >> >>   chartbl[0x2a] = 'á';
> >> >> >
> >> >> > As a matter of curiosity: what compiler and environment?
> >> >> >
> >> >> >> --- sub/sub_cc.c	(revision 34760)
> >> >> >> +++ sub/sub_cc.c	(working copy)
> >> >> >> @@ -65,16 +65,16 @@
> >> >> >>    for (i = 0; i < 128; i++)
> >> >> >>      chartbl[i] = (char) i;
> >> >> >>    /* now the special codes */
> >> >> >> -  chartbl[0x2a] = 'á';
> >> >> >> -  chartbl[0x5c] = 'é';
> >> >> >> -  chartbl[0x5e] = 'í';
> >> >> >> -  chartbl[0x5f] = 'ó';
> >> >> >> -  chartbl[0x60] = 'ú';
> >> >> >> -  chartbl[0x7b] = 'ç';
> >> >> >> -  chartbl[0x7c] = '÷';
> >> >> >> -  chartbl[0x7d] = 'Ñ';
> >> >> >> -  chartbl[0x7e] = 'ñ';
> >> >> >> -  chartbl[0x7f] = '¤';    /* FIXME: this should be a solid block */
> >> >> >> +  chartbl[0x2a] = 0xe1;
> >> >> >> +  chartbl[0x5c] = 0xe9;
> >> >> >> +  chartbl[0x5e] = 0xed;
> >> >> >> +  chartbl[0x5f] = 0xf3;
> >> >> >> +  chartbl[0x60] = 0xfa;
> >> >> >> +  chartbl[0x7b] = 0xe7;
> >> >> >> +  chartbl[0x7c] = 0xf7;
> >> >> >> +  chartbl[0x7d] = 0xd1;
> >> >> >> +  chartbl[0x7e] = 0xf1;
> >> >> >> +  chartbl[0x7f] = 0xa4;    /* FIXME: this should be a solid block
> >> >> >> */
> >> >> >>  }
> >> >> >
> >> >> > Assuming the original code was right and supposed to be ISO-8859-1,
> >> >> > the
> >> >> > change seems fine. I would like it even better with:
> >> >> >
> >> >> > +  chartbl[0x2a] = 0xe1; /* á */
> >> >> > +  chartbl[0x5c] = 0xe9; /* é */
> >> >> > +  chartbl[0x5e] = 0xed; /* í */
> >> >> > +  chartbl[0x5f] = 0xf3; /* ó */
> >> >> > +  chartbl[0x60] = 0xfa; /* ú */
> >> >> > +  chartbl[0x7b] = 0xe7; /* ç */
> >> >> > +  chartbl[0x7c] = 0xf7; /* ÷ */
> >> >> > +  chartbl[0x7d] = 0xd1; /* Ñ */
> >> >> > +  chartbl[0x7e] = 0xf1; /* ñ */
> >> >> > +  chartbl[0x7f] = 0xa4; /* ¤ FIXME: this should be a solid block */
> >> >> >
> >> >> > The compiler should not choke on non-ASCII in comments.
> >> >>
> >> >> We are better without them.
> >> >> They could still cause issues, when viewing/editing the file on UTF-8
> >> >> locale.
> >> >
> >> > Just make the characters UTF-8.
> >>
> >> Then it would break on all single-byte codepage locales. The utf8
> >> sequence would be interpreted as multiple characters, that is not
> >> allowed in a literal.
> >
> > I'm talking about the comments.
> 
> I would really like to avoid having any non-ascii characters in source
> files, unless absolutely necessary.
> It also may be a little bit confusing to have utf8 encoded characters
> of latin1 codes.

Too late, we have UTF-8 all over the place.

> I won't oppose adding such comments as a second patch.

Now that would be nonsensical clutter.

Diego