Re: Translations, GNOME and KDE

[Danilo Segan <danilo gnome org>]

Luis Villa <louie ximian com> writes:

> On Sun, 2004-01-04 at 11:45, Jeff Waugh wrote:
>> > 2-Reduce the pos to only UI visable strings, no normal user will take a
>> > look at Debug strings and so speak with coders about which strings to
>> > include on the pos and which not...
>> 
>> Possibly controversial, but not something I'll get into. ;-)
>
> FWIW, I don't think this should be controversial at all. Coders who
> don't know English are already in a deep hole, and debug string in a
> foreign language reported to english-speaking hackers or (worse) into
> bugzilla are of virtually no use. If it goes to console, it should never
> be marked for translation (IMHO.)

This is too simple-minded, IMHO.  I consider myself a technical
person, and yet I prefer to see every output in my own language.  I
even have the habbit of naming variables, functions and such in my
own language, write comments in my own language etc.

Being a GNOME developer (I'm not one -- at this time ;) means
accepting a *policy* of English being the default: not every coder
actually prefers it for writing code. There are obvious advantages of
using English (easier to share), but it has nothing to do with debug
strings being translatable: if one can read/write code, and gets a
debug message she understands, she might help sooner rather than if
she gets a message she doesn't grasp because it's in English.

Good examples are the recently removed gcalctool messages. One of the
messages was of the form "Error: Newton's iteration doesn't
converge" (or something similar).  This is a place where a
mathematican might help if she understands where the problem is.  Any
programming voodoo is usually withing grasp of many mathematicians,
and being able to understand the message, no matter how "debuggy" it
is, might prove invaluable.

Of course, this is purely imaginary, and probably quite rare case, so
it all reduces to "My *preference* is to have debug messages
translated."  This needs to be weighted against "no unneccessary
workload for translators".

(FWIW, in the Linux kernel 2.2 era, I've even translated some of the
printk messages just for fun -- I believe I even saw a project to do
that systematically not long ago; though, from the desktop stance,
kernel is a technology and the way of its operation is not really
relevant -- my main point while doing that was: when something screws
up badly, that's when you most need to understand what has happened)


Cheers,
Danilo