Hspell plugin for KSpell2 seems fine now

[Zack Rusin]

On Friday 29 April 2005 12:21, Mashrab Kuvatov wrote:
> > You may notice, though, that in Hebrew a legal word may contain the
> > quote and the double quote sign. It may be difficult, but it would
> > be better Hebrew-wise if you don't break words on these characters.
> >
> > For example, you have the word ה"א which is the perfectly legal
> > name of the fifth letter of the hebrew alphabet. I believe that if
> > you pass it as a single word, the red wave below it would
> > disappear.
>
> This will require some modifications in KSpell2. It is worth to try.

The current algorithm for breaking up the words is really bad. It barely 
works for western languages. It was more a proof of concept for the 
idea of a context dependent feeder in KSpell2 that could be subclassed.

The concept of separating text into entities (and I'm purposly not using 
the word "words") that can be spellchecked is not trivial (very 
interesting though) for almost anything besides western languages. If 
anyone does have a good paper on the topic I'd love to read it. 

> > Thanks again for your integration work, even though it won't be
> > checked-in until we abide to your (sensible, I must agree)
> > ultimatum regarding word addition in runtime.
>
> Oh, no it is not ultimatum, it is just my personal opinion. Do not
> get me wrong. Zack is author and maintainer of the KSpell2, he can
> add it into KSpell2 anytime he wants.

First of all thank you for working on this. Before commiting it, could 
you write a configure check for hspell and conditionally compile the 
plugin if it's present? (it's basically the same code that was written 
for aspell plugin). Oh, and setting runTogether to false again in the 
filter is uncessary, it already is false.

Zack

-- 
15% of the populace will never steal. 5% of the populace will steal 
almost anything not nailed down.