Re: [bug-gettext] [Translation-team-de] Swiss typography

[Vladimir 'φ-coder/phcoder' Serbinenko]

On 06.03.2012 17:46, Andreas Stricker wrote:
> 06.03.12 17:04, Vladimir 'φ-coder/phcoder' Serbinenko wrote:
> > On 06.03.2012 16:52, Andreas Stricker wrote:
> > > 04.03.12 17:17, Vladimir 'φ-coder/phcoder' Serbinenko wrote:
> > > > I have found some more differences/lines for the sed:
> > > > terminology usage (fortunately they are of the same gender, so no need
> > > > to more difficult grammar adaptation):
> > > > s,Rechner,Computer,g
> > > > Keyboard difference:
> > > > s,Strg,Ctrl,g
> > > > s,Einfg,Insert,g
> > > > s,Entf,Del,g
> > > This looks dangerous for me. At least the last one will render
> > > a sentence like
> > >
> > > "Entfernen Sie nun die DVD aus dem Laufwerk"
> > >
> > > into
> > >
> > > "Delernen Sie nun die DVD aus dem Laufwerk"
> > >
> > > Certainly not an expected behavior...
> > What about:
> > s,ß,ss,g
> > y,»«,«»,
> Those are not that risky, because this are special characters
> only appearing in context of a German source text.
>
> > s,\([^a-zA-ZöüäÖÜÄ]\|^\)Rechner\([^a-zA-ZöüäÖÜÄ]\|$\),\1Computer\2,g
> > s,\([^a-zA-ZöüäÖÜÄ]\|^\)Strg\([^a-zA-ZöüäÖÜÄ]\|$\),\1Ctrl\2,g
> > s,\([^a-zA-ZöüäÖÜÄ]\|^\)Einfg\([^a-zA-ZöüäÖÜÄ]\|$\),\1Insert\2,g
> > s,\([^a-zA-ZöüäÖÜÄ]\|^\)Entf\([^a-zA-ZöüäÖÜÄ]\|$\),\1Del\2,g
> Use the word boundary expression "\b"
>
> s,\bRechner\b,Computer,g
> s,\bStrg\b,Ctrl,g
> s,\bEinfg\b,Insert,g
> s,\bEntf\b,Del,g
>
> Continuing constructing automated rules quickly gets tricky and either
> has false-positives or false-negatives.
false negatives are fine. There is no goal of eliminating every 
reference to German standard, since people here usually have no problem 
understanding those (perhaps after couple of seconds thinking).  It's 
more about making text easier to grasp for Swiss users.
>   When replacing words you quickly
> end up in situations where a full understanding of the context is
> required.
>
> You may provide such automated rules to create an initial translation.
> But then a human should proof read it afterwards.
This goes into the question of necessary resources compared to the 
importance of the goal. As explained before it's not that important to 
adjust every occurrence. IMHO the cases which can't be reliably handled 
automatically just aren't worth the hassle. If someone disagrees and is 
willing to provide the work needed to proofread the unreliable cases the 
proofread approach can be dicussed but otherwise it's pointless.
> Regards, Andy


--
Regards
Vladimir 'φ-coder/phcoder' Serbinenko