On Mon, 9 Mar 2009 07:53:12 +0000 (UTC) Guenter Milde <mi...@users.berlios.de> wrote:
> On 2009-03-07, Micha Feigin wrote: > > > ... I was wondering why lyx uses it's own keyboard switching, > > especially for hebrew. Is it a technological issue or just that no one > > had the time/interest to implement this yet? > > * LyX does not necessariyl use its own keyboard switching but provides > this as an alternative. > > * One LyX developer preferred to keep the rest of his system in English > when writing Hebrew in LyX. This is why he insisted on keeping the > parallel feature. > > > Is it even possible to know what language is chosen on all the systems > > lyx is implemented on > > LyX recognizes the "locale" setting and uses it to "speak to you" in > native tongue (if possible) -- my LyX speaks German. > I don't care what language it speaks, that is trivial, it's a question of changing the writing language inside the paragraph or between paragraphs. > It would also be possible to use this to set the document language for > new documents. However: > > > or does the language need to be guessed based on each character? > > it is not possible (in general) to determine the language from a > character: If LyX sees an 'ü', should this be German, Hungarian, or > Turkish. And how about "Häagen Dasz" or "TεX"? No language switch needed. > > (The case is somewhat simplified with Greek or Hebrew characters, but > even then you dont know whether it is Hebrew, New-Hebrew or Yiddish.) > There are two issues. For running the dictionary you need to know the language. For hebrew and arabic it's another issue, you need to know the system language so that you know directionality. Hebrew is right to left. For hebrew characters it may be easy to decide, for what about spaces and numbers? For these we need to know the system keyboard language and not guess it from the character. Under windows I know it's possible since for example word does it. Question is whether this is possible to know under linux (I guess so since there are panel applets that show the language). Which again comes down to the question whether there is a technical issue why to work this way or not. > But the problematic cases were ASCII chars in a Greek or Heberew > document: the choice of alternative language is just too big to give a > sensible guess. > there is another question here, what about the correct encoding ... > Günter > >
