Re: Better error message ?

[Bruno Haible]

Felix Natter writes:

> #: gtypist.c:1470
> #, c-format
> msgid "'%' must immediately follow value"
> msgstr "«%» debe seguir un valor"
> 
> I get this error:
> 
> rm -f es.gmo && /usr/local/bin/msgfmt -c --statistics -o es.gmo es.po
> es.po:76: 'msgstr' is not a valid C format string, unlike 'msgid'
> /usr/local/bin/msgfmt: found 1 fatal error
> 
> This is misleading because 'msgid' is not a valid format-string

"'%' must immediately follow value" is a valid C format string. The
format directive starts at '%' and ends at 'm' (this is a glibc
extension).

> either. I think it should better be something like (I hope you have a better
> idea):
> es.po:76: 'msgstr' is not a valid C format string as the comment suggests
> or
> es.po:76: 'msgstr' is not a valid C format string but it is marked as one

Such an error message would suggest to the translator to remove the
"#, c-format" mark, which won't help. (The mark would be reintroduced
when the maintainer runs msgmerge.)

> I also have a question concerning the code:
> the code looks like this:
> 
>         fatal_error( _("'*' must immediately follow '%'"), copy_of_line );
>       else
>         fatal_error( _("missing '%'"), copy_of_line );
>       } else
>       fatal_error( _("missing '%'"), copy_of_line );
>     }
>     if (isspace( *(data - 1) ))
>       fatal_error( _("'%' must immediately follow value"), copy_of_line );

In order not to let xgettext think these are format strings, it'd be
good to mark these strings as being not C printf strings.


          /* xgettext: no-c-format */
          fatal_error( _("'*' must immediately follow '%'"), copy_of_line );
        else
          fatal_error( _("missing '%'"), copy_of_line );
      } else
        fatal_error( _("missing '%'"), copy_of_line );
    }
    if (isspace( *(data - 1) ))
      /* xgettext: no-c-format */
      fatal_error( _("'%' must immediately follow value"), copy_of_line );

Bruno