Help changes, problematic for l10n

Hi, all,

today I filed two bugs regarding recents changes in help + tooltips (for
LO7) that might cause problems to other l10n teams (other than Slovenian):
https://bugs.documentfoundation.org/show_bug.cgi?id=130979
https://bugs.documentfoundation.org/show_bug.cgi?id=130993

Please feel free to comment.

Lp, m.

I introduced %MOD1 and %MOD2 for the tip-of-the-day strings to be specific on Windows/Linux vs. macOS. You can translate, for example, STR_CMD defaulting to "⌘ Cmd" to something else.

Martin asks in tdf#13079 to revert this patch and to duplicate the full strings in order to allow cases. Sounds like the wrong way to me and I would like to get more input first. One point is that I cannot find any example that requires a case (in German I can always talk about "%MODx-key" and apply the case to "key").

The full list of strings is here https://opengrok.libreoffice.org/xref/core/cui/inc/tipoftheday.hrc

TIA,
Heiko

PS: @Martin: many thanks for looking very closely on every detail. I really appreciate that even when disagreeing in this particular situation.

Hi, Heiko,

the system you introduced does not allow grammatical cases of the localized
keys. So it doesn't work OK for some languages, it is faulty.

Of course, if you think it is possible, you can upgrade it to work, so it
will have also a parameter in which case the string should be, and also the
build process should then use the proper dictionaries for all l10n
languages of LO (possible to generate all word cases), so that could work.
It seems a complex job to me, but I might be wrong.

Lp, m.

V V pet., 28. feb. 2020 ob 08:20 je oseba Heiko Tietze <
heiko.tietze@documentfoundation.org> napisala:

Hi Heiko, Martin, *,

Martin Srebotnjak (<miles@filmsi.net>), 28.02.2020 kell 12:49:

Hi, Heiko,

the system you introduced does not allow grammatical cases of the localized
keys. So it doesn't work OK for some languages, it is faulty.

Of course, if you think it is possible, you can upgrade it to work, so it
will have also a parameter in which case the string should be, and also the
build process should then use the proper dictionaries for all l10n
languages of LO (possible to generate all word cases), so that could work.
It seems a complex job to me, but I might be wrong.

Another thing to consider would be small vs. capital first letter, for the
languages that don't always use the capitalized version: "Ctrl/Alt/Option"
are (presumably) always written with capital initial in the languages where
they are named as in English (such as in my native Estonian), however the
Slovenian "krmilka" (note the small first letter) would then need to be
somehow capitalized when it's used in the beginning of a sentence.
*Possibly* this situation can be avoided, *if* the localizers always
remember that "strilka"/"strilke" etc. will always appear in all-lowercase,
and *if* the language's grammar allows for playing around with word order -
but that's two big "if's".

I'm not sure it's worth the trouble. More failure-proof option would be to
just have separate strings for Linux/Windows and Mac. Sure, it means a bit
more work for l10n teams, but translation memory can alleviate that.

Best,
Mihkel
Estonian translator