Re: Problem in documentation: number of plurals

[Gabor Kelemen]

Hi

I second that all. 
Furthermore, the gettext-tools/src/plural-table.c file, which also 
contains this information, needs to be patched (see attachment), so that 
msginit will return correct information.

The problem is serious, because some translators, like me, used Kbabel  
for translation, which used msginit to guess the correct plural forms. 
Thus, there are lots of Hungarian translations around with incorrect 
plurals information, as Kbabel allowed to use two plurals in 
translation, while putting nplurals=1; plural=0 in the header (seems to 
be another bug).
This worked fine until gettext 0.15, as gettext didn't cared with 
differences between the nplurals information in header and the number of 
actually used plurals. With Gettext 0.15, this behavior changed and some 
translations [1] already broke the build process of the respective package.

KDE translations are another kind of problem, they use the format 
described below to handle plurals, and they are using msginit (or at 
least the same information msginit uses, i'm not familiar with KDE) 
real-time to determine the number of plurals to search for in the 
translation files. Consequently, they find mismatch between the numbers 
of expected and found plurals, and display an ugly error message instead 
of the string. In this case, my patched version of Gettext solved this 
problem instantly.

So i'd like to ask you to apply the attached patch and roll out a new 
release, in order to make sure Kbabel users will not create more build 
process breaker translations, and also to make KDE-style translations 
work correctly.


Regards

Gabor Kelemen

[1]: 
http://sourceforge.net/mailarchive/forum.php?thread_id=29984621&forum_id=8113

Arpad Biro írta:
> Hi,
>
> Section "10.2.5 Additional functions for plural forms" of the
> gettext documentation ( http://www.gnu.org/software/gettext/manual )
> claims:
>
>   Only one form:
>
>     Some languages only require one single form. There is no
>     distinction between the singular and plural form. An appropriate
>     header entry would look like this:
>
>       Plural-Forms: nplurals=1; plural=0;
>
>     Languages with this property include:
>
>     Finno-Ugric family
>         Hungarian
>
>
> There's a problem with this: it's NOT true when there's no
> numeral involved.
>
> A quick example:
>
> "1   apple "   is   "1   alma"   in Hungarian, and
> "123 apples"   is   "123 alma"
>
> So far so good, but when there's no numeral:
>
> "the apple "   is   "az alma "   in Hungarian, but
> "the apples"   is   "az almák"
>
> In the latter case we have 2 plural forms.
>
>
> An example from the Gwenview POT file:
> (
> http://websvn.kde.org/trunk/l10n/templates/messages/extragear-graphics/gwenview.pot
> )
>
> #: gvcore/deletedialog.cpp:97
> msgid ""
> "_n: <qt>This item will be moved to the trash bin.</qt>\n"
> "<qt>These items will be moved to the trash bin.</qt>"
>
>
> This has to be translated into 2 different forms in Hungarian
> (just like it is in English).
>
>
>
>
>
> Arpad Biro
>
>
> __________________________________________________
> Do You Yahoo!?
> Tired of spam?  Yahoo! Mail has the best spam protection around 
> http://mail.yahoo.com 
>
>
> _______________________________________________
> address@hidden
> http://lists.gnu.org/mailman/listinfo/bug-gnu-utils
>
>
>
>   



--- /home/gabor/gettext-orig/gettext-tools/src/plural-table.c 
+++ /home/gabor/gettext-0.15/gettext-tools/src/plural-table.c 
@@ -26,7 +26,6 @@
 /* Formulas taken from the documentation, node "Plural forms".  */
 struct plural_table_entry plural_table[] =
   {
-    { "hu", "Hungarian",         "nplurals=1; plural=0;" },
     { "ja", "Japanese",          "nplurals=1; plural=0;" },
     { "ko", "Korean",            "nplurals=1; plural=0;" },
     { "vi", "Vietnamese",        "nplurals=1; plural=0;" },
@@ -36,6 +35,7 @@
     { "en", "English",           "nplurals=2; plural=(n != 1);" },
     { "fo", "Faroese",           "nplurals=2; plural=(n != 1);" },
     { "de", "German",            "nplurals=2; plural=(n != 1);" },
+    { "hu", "Hungarian",         "nplurals=2; plural=(n != 1);" },
     { "nb", "Norwegian Bokmal",  "nplurals=2; plural=(n != 1);" },
     { "no", "Norwegian",         "nplurals=2; plural=(n != 1);" },
     { "nn", "Norwegian Nynorsk", "nplurals=2; plural=(n != 1);" },