[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Re: build it vs. adapt it
- To: Development Discussions <developer at arabeyes dot org>
- Subject: Re: build it vs. adapt it
- From: Behdad Esfahbod <behdad at cs dot toronto dot edu>
- Date: Thu, 7 Apr 2005 16:06:30 -0400 (EDT)
On Thu, 7 Apr 2005, Nadim Shaikli wrote:
> Behdad, alot of what we do will sooner or later make it into the
> more mainstream applications (consider them proof-of-concept and
> intent as well as clear indicator of seriousness about the topic).
> Alot of "our" Arabic-centric projects, as you note, were started
> after lengthy conversations with the various developers (such as
> duali - the spell-checker) to where Arabic inclusion was VERY
> unlikely -- aspell, back then, was in that boat. And even in the
> case of duali we've had a number of conversations here and elsewhere
> about getting it adapted (I'm unaware of aspell's current status
> and abilities and so I don't know if it's easier now or harder to
> include Arabic).
>
> Our collective intent has always been to contribute upstream to
> the more popular applications and that will happen, but alot of
> work can take place without the consent and/or approval of the
> upstream application to flush out various ideas/algorithms.
> Arabeyes' about [1] page clearly notes this as well,
>
> "Arabeyes' goal is not to create new applications (if those
> applications exist), but to incorporate modifications/additions
> to existing common everyday-use applications."
>
> Hope that makes sense - in short, we fully agree.
>
> Salam.
>
> - Nadim
All these makes sense, but a few points:
* You don't have to wait for aspell to prepare itself for
Arabic to be added, you can add the infrastructure too.
* You don't have to wait for upstream maintainers, just import
the code in Arabeyes CVS and go from there. When you have
something useful out of it, send patch upstream. If you've done
it right, it would be accepted in a silver platter.
* In the case of OCR, I repeat myself again, a lot of useful
code is in GOCR, that unless you use, you are going to rewrite
here. Like I said, TIFF handling code is out of place in an OCR
application. And it makes integration harder later on too.
* Arabeyes stated policy is ideal, just that I like to see it
more applied.
Cheers,
--behdad
http://behdad.org/