Janis,

Yes, I agree with that.
Also for other languages we need a mechanism to 
limit the prefix-suffix combinations. The Y/N mechanism 
is too coarse.

I think, to keep the groups apart separation works fine, like:
dazugehÃren/Db
dazugehÃren/IXY
This makes sure, that the endings generated by the 
IXY flags do not get paired with the prefixes generated
by the b prefix. These combinations would be grammatically
nonsense.

The other way create prefix classes is the 
secondary flag in the Hungarian spell checking,
described under 
http://tkltrans.sourceforge.net/tklspell/secflag.htm

But I think, it would be good, if Kevin fixed the
error with the prefixes first.
http://www.openoffice.org/issues/show_bug.cgi?id=44220

Regards, Eleonora

Am Sonntag, 6. MÃrz 2005 20:54 schrieb Jancs:
> CitÄju "Kevin B. Hendricks" <[EMAIL PROTECTED]>:
> > Hi,
> >
> > > dic contains:
> > > dazugehÃren/Db
> > > dazugehÃren/IXY
> >
> > This is not allowed for in MySpell.  All root words in the dictionary
> > must be unique.  This is always possible by simply expanding one of the
> > two sets of affixes if for some reason they can't be merged into one
> > entry just by merging  affixes.
>
> And in the same time, especially in cases when by some disastrous thoght
> you decided to use the PFX keys also, munch is giving you some kind of well
> organized trash: under some circumstances it decides that the _form_ of the
> root word with PFX keys must be presented along with the root word having
> full PFX/SFX asset. This often ends with many unnecessary dic entries which
> one have to hunt down by grep-ing them out.
>
> It was described somewhere. But i as a non-linguist can not get the point
> of why to PFX is given the same priority as to SFX? May be someone could
> give me explanation as Mr. Hendricks is not answering on my questions.
>
> P.S. I can agree that the aff file of lv_LV is not designed in the way
> which corresponds to the MySpell idea, but it works. Here are still
> questions, for example - what to do in cases when 3 different root words in
> three different declinations are giving the same form. May be re-desing
> everything based on the roots of the words and not the root-words which by
> my understanding is the basic form of the word, something like courir (Fr,
> which with its horrors of grammar can be comapred with Lv).
>
> Janis

---------------------------------------------------------------------
To unsubscribe, e-mail: [EMAIL PROTECTED]
For additional commands, e-mail: [EMAIL PROTECTED]

Reply via email to