Re: Code for peer-review: SpellChecker class

Subject: Re: Code for peer-review: SpellChecker class
From: Dom Lachowicz (
Date: Wed Apr 11 2001 - 08:20:29 CDT

Hi Tomas,

>(1) should not the parameter be const XML_Char * ?

No, it shouldn't. I prefer to only use XML_Char for XML importing/exporting
type stuff.

>(2) I would suggest that both for Ispell and Pspell, the szLang
>refers to the same tag of the format 'en-US', and that the translation
>to the ispell hash path is done within the ispell implementation of
>the SpelChecker class. This will make the integration into the
>existing code easier and the virtual class more virtual.

Agreed. This'd be great actually.

>(3) I am not sure I understand what is happening in the function
>itself. I think we should first test whether m_map already contains
>our language and if so return the associated SpellChecker, and
>only if not try to allocate a new SpellChecker. I also suspect this
>function will have to look slightly different for Pspell and Ispell, for if
>I understand correctly, we will need only one instance of Pspell
>checker which handles multiple languages, while with Ispell we will
>need an instance for each language (which also means that the in
>the SpellManager destructor, we will have to delete these multiple
>instances as well).

Oversight. In my caching code I forgot to actually cache :-)

Thanks for all of the comments both on and off-list. The next rev is done
and I think that I'll start implementing the changes in the source tree.

I'll need someone to do the legwork for ispell though because I plan on
breaking that build to get this to work. Volunteers?


Get your FREE download of MSN Explorer at

This archive was generated by hypermail 2b25 : Wed Apr 11 2001 - 08:20:33 CDT