Re: an i18n sample

["Eli Zaretskii" <eliz at gnu dot org>]

> Date: Tue, 26 Oct 2004 16:01:56 -0400
> From: Daniel Jacobowitz <drow@false.org>
> Cc: ibr@ata.cs.hun.edu.tr, gdb-patches@sources.redhat.com
> 
> I mean that I believe the string that GDB displays to the user for
> these two cases, i.e. an unnamed type or a type whose name is set to
> <unnamed>, should always be the same.

If that is possible in the target language, I agree that the strings
should be similar.  Having 2 complete strings will help in getting as
close to this goal as possible, since the translator will see two
similar sentences one after the other in the message catalog and
figure out that they are from the same code.

> > Given a string "<unnamed>", how would a translator know that it is
> > supposed to be a substitution for %s in the format string?  The only
> > way to know that is to read the source, which a translator normally
> > does not do.  Without knowing the context of "<unnamed>", the
> > translator is unlikely to find a good translation for it.
> 
> That's true.  But it expresses a concept - an object without a name.
> Would "<unnamed type>" be better?

It's possible, I don't know.  My concerns were about the linguistic
and grammatical aspects: for example, in some languages, adjectives
such as "unnamed" have different forms depending on whether they refer
to a noun of a masculine or feminine (or even neuter) gender.
"Unnamed type" will actually help in this case, but other similar
grammatical considerations might still remain unresolved.  Having a
full sentence normally goes a long way towards resolving them.