Re: [Bug-gnulib] Re: version-etc, AUTHORS, WRITTEN_BY

[Bruno Haible]

Paul Eggert wrote:
> My main worry here is that author's names will have to be translated
> differently in different contexts, and that by removing the context
> we're making translation harder.
>
> For example, suppose the proper translation for "Written by Paul
> Eggert." is
>
> Eggert Paul-er do desu ka.
>
> where "Eggert Paul-er" is the translation of "Paul Eggert", and the
> "-er" suffix differs for different names.

Ehm, which language are you talking about? IMO the principle of a proper
name is that it can serve to identify a person, _regardless_ of context.
This is the primary reason for a proper name to be written in a passport,
green card etc., isn't it?

I'm not aware of any language where it is _mandatory_ to prefix or suffix
proper names. We all know that the polite Japanese form of your name is
Eggert-san, but it is also ok without the suffix.

In GNU gettext, we've been using the "Written by %s.\n" string for 8 years
now, with 28 different languages, and no translator or user has complained
about the result of substituting "Ulrich Drepper" for "%s".


> Perhaps I'm wrong, but I suspect that
> translators would prefer translating whole sentences to translating
> bits and pieces.

Yes, translators prefer to translate whole sentences when there _are_
variations. For example,

    _("Written by %s, %s, and %s.")

is preferred over

    _("Written by %s."), %s = _("%s, %s, and %s")

But when there are none, like here (from GNU awk),

   #: awkgram.y:232
   msgid "BEGIN blocks must have an action part"
   msgstr "Jeder BEGIN-Block muss einen Aktions-Teil haben."

   #: awkgram.y:238
   msgid "END blocks must have an action part"
   msgstr "Jeder END-Block muss einen Aktions-Teil haben."

the translator prefers a single message to multiple ones.

The ice in thin here, I admit. It's only the passive voice of "Written by"
which saves us. If the English message were "Author: %s\n"
then the italian translation ("Autore: %s\n" or "Autora: %s\n") would
depend on the gender of the name.


> I merely think the current
> proposal is over-engineering that in the long run will hassle
> developers and translators more than the simple approach would.

The experiment of "Written by %s.\n" in GNU gettext lasts for 8 years,
which is a "long run", and there are no hassles associated with it.

Bruno