[PATCH] KSpell2 improvement

[Nicolas Goutte]

On Monday 13 March 2006 15:01, Mashrab Kuvatov wrote:
> On Monday 13 March 2006 10:06, Thiago Macieira wrote:
> > Finally, "tet" is Tetum. But I'm not sure we have the 3-alpha
> > translations mapped.

KDE3 uses 2 letters if possible, if not it uses 3 letters (and then RFC 3066 
but with _ instead of - as separator).

>
> nds (Low Saxon) and hsb (Upper Sorbian) are treated properly.
>

> Is anybody going to add no, tl and tet?

If you mean in kdelibs/kdecore/all_language.desktop, I would not mind for new 
languages (2 letter code is prefered, 3 letter code otherwise), except that 
this is really close to the next KDE release, which is this week.

However for Norwegian, I do not think that "no" should be added, as KDE makes 
a distinction between the two main Norwegian variants.

>
> > >One todo is left. One has to translate dictionary cryptic extension
> > > names into English and i18n'ify it. I have no idea how to do it.
> >
> > I'd recommend we use the form "Language (Region) Variant" for English and
> > construct it as
> > 	i18n("%1 (%2) %3").arg(i18n(language), i18n(region), i18n(variant))
> > so that translators get the chance to translate everything tokens.
>
> One doesn't need i18n for language and country. Like I said they are
> already translated.
>
> > Do not construct the strings like you did:
> > +              localizedDictionary += " (" + cISOName + ")";
>
> Why?

Because you cannot always do in any language what you can do in English. What 
might be "Language (Region) Variant" in English might need to be translated 
somewhat like "Language (of Region) of Variant" with "of" not being a 
separate word but a declinaison of the word "Region" or "Variant".

>
> > That means: extract all the extensions known to aspell into a file, map
> > them to an English name and mark them for translation.
>
> I've never done this so far and I don't know how the messages are
> extracted. If somebody helps me (Nicolas Goutte?), I'll try. Otherwise I
> leave it to experts.

Well, the goal is to create a dummy C++ file with entries like:
i18n( "Language (Region) Variant" );

And then in the real code, you use such a tring as base for the translation.

>
> > >Would be nice to get it into stable (as it is there are no new
> > > messages). KWord and KPresenter users would appreciate it. By the way,
> > > one can do the same for dictionary selection combo in KWord's
> > > preferences for example.
> >
> > We're in feature-freeze. This is KDE 4 code...
>
> Personally, I don't take it as a feature. For me it is a bug and it is
> fixed without introducing any new messages.
>
> Like I mentioned, at the moment KOffice apps are using kspell2 (maybe other
> apps too, I'm not aware of it). So, if KDE thinks that users should wait
> for KOffice 2 to get a change above in a spellchecker dialog, then let it
> be.

If KDE 3.5.2 is too late, which I really think it is, there is still KDE 
3.5.3.

>
(...)
>
> Cheers,
> Mashrab.

Have a nice day!