Re: Build failures on Ubuntu due to gettext

[Colin Watson]

On Tue, Dec 08, 2009 at 12:59:28AM +0100, Vladimir 'φ-coder/phcoder' Serbinenko 
wrote:
> Colin Watson wrote:
> > Furthermore, the style I suggested is used in many other GNU projects.
> > Here are just a few examples:
> >
> >   coreutils/src/du.c:994:       error (0, 0, "%s", _("invalid zero-length 
> > file name"));
> >   diffutils/src/sdiff.c:853:  fprintf (stderr, "%s", _("\
> >   diffutils/src/sdiff.c-854-ed:\tEdit then use both versions, each 
> > decorated with a header.\n\
> >   glibc/inet/rcmd.c:178:                              __fxprintf(NULL, 
> > "%s", _("\
> >   glibc/inet/rcmd.c-179-rcmd: socket: All ports in use\n"));
> >   gnulib/lib/xalloc-die.c:34:  error (exit_failure, 0, "%s", _("memory 
> > exhausted"));
> 
> I have nothing against defining grub_putl_ (str); equivalent to
> grub_printf ("%s\n", str); (defined as function if it's used extensively
> for space reasons).
> Just from your previous mails it seemed that you proposed to transform
> strings like
> grub_printf (_("Moving file %s to %s."), f1, f2); to grub_printf ("%s %s
> %s %s.", _("Moving file"), f1, _("to"), f2);
> to avoid translating format-strings which would be completely
> untranslatable.

What on earth?! I said nothing of the kind! That would obviously be
completely insane. How did you arrive at that impression?

My patch made the following transformation:

-  grub_printf (_("literal string"));
+  grub_printf ("%s", _("literal string"));

This was only necessary in five places, so I doubt that a function is
worth it.

Thanks,

-- 
Colin Watson                                       address@hidden