[wp-polyglots] 2.3 Beta 1 & strings descriptions

[F Wolff]

Op Dinsdag 2007-08-28 skryf Nikolay Bachiyski:
> Hello fellow translators,
> 
> As you may all know, Beta 1 is now out [0]. You can find its pot at
> the usual place [1].
> 
> I want also to give you some information about a new feature of our
> i18n system. Sometimes there are strings, which are completely
> incomprehensible for translators or are used several times in
> different contexts.

Please excuse me if my comments are stupid - perhaps I missed some
earlier discussion about this.

> 
> Here is an example of an incomprehensible string:
> %1$s%2$s%3$s
> You just don't know what the hell it means. The new approach is to
> append a vertical bar and string clarification after the string, so
> don't be surprised if you see a string like:
> %1$s%2$s%3$s|1: Pages, 2: by %s, 3: matching %s
> Thus you can both reorder the three strings and actually know what you
> are reordering. Handy, isn't it?

Not really. I probably can't use TM matching (not that relevant in this
case), or any automated quality assurance on this translation. In fact,
if TM matching is used, it might actually match on the part that I
shouldn't be translating. Why don't the developers use programmer
comments (also called automatic #. style comments) like other PO based
projects? xgettext will usually put the comments in the source files
above the gettext call into the #. comments of the POT file
automatically. Alternatively you can use an option for xgettext:
	-c, --add-comments[=TAG]
              place comment block with TAG (or those preceding keyword  lines)
              in output file

For the specific example you mention, why isn't the message just one
long message with everything in it? Wouldn't this kind of substitution
break in languages with declension anyway?

> Now, let's see the different contexts case:
> »
> Very common string, which is usually used as a button hint. However in
> this case, it is used as a list bullet and, especially regarding
> right-to-left languages, might need different treatment. The current
> string for that now is:
> »|Used as a list bullet

If the issue is that the same text is used in different contexts, isn't
the solution to use msgctxt? This ensures proper opportunities for QA,
fuzzy matching, reuse, etc.

Perhaps just as useful in this case is to tell translators in #.
comments what "raquo" actually is. I can't imagine that everybody knows
that.

If we really, really need to stuff the msgid with context information,
why don't we at least use one of the existing schemes like the ones used
before in KDE? At least some translation tools tried to support it.

Keep well
Friedel