Subject: Re: comparison -- locales with ispell dictionaries
From: Vlad Harchev (hvv@hippo.ru)
Date: Fri Mar 02 2001 - 00:24:17 CST
On Thu, 1 Mar 2001, WJCarpenter wrote:
> Something I have been curious about ever since I wandered into the
> ispell code in AbiWord is... how's come ispell is compiled into the
> application instead of being run as a separate process?
>
> The conversation the app needs to have with the spell-checker is
> pretty simple, so the subprocess app can be pretty simple. The
> advantage to this approach is that it is simpler to support different
> spell checking phenomena different places, including using many of the
> different variants of ispell found around the universe with slightly
> different dictionary formats.
>
> It also probably makes it architecturally simpler to do things like
> run spell checking for multiple locales in a document simultaneously.
> I admit I'm talking through my hat on this one. It may be quite
> simple with Pspell, but I covered my ears when I heard that talk about
> Pspell dynamically loading shared libraries to get things done.
>
> I've wondered about this because my close personal friend, emacs, has
> used a spell-checking subprocess for hundreds, perhaps thousands, of
> years.
Hmm, as I understand, aspell is an advanced spell checker that learns what
mistakes user makes and suggests most probable replacements. So, the
information about user mistakes should aggregate while the AW is running. This
means that either that pspell (an interface to pspell and aspel) is compiled
into the AW, or these spell checkers are used via pipes. Due to portability
issues pipes may not be an option (I'm not absolutely sure though) - so the
only choice is compiling in, alas.
Pspell has problems if used with non-gcc compilers since it needs to use
C++ vtable from shared library. IIRC I recall, only gcc supports this, and
probably only on unix.
> --
> bill@carpenter.ORG (WJCarpenter) PGP 0x91865119
> 38 95 1B 69 C9 C6 3D 25 73 46 32 04 69 D6 ED F3
>
Best regards,
-Vlad
This archive was generated by hypermail 2b25 : Fri Mar 02 2001 - 01:00:47 CST