Re: [aspell-devel] Aspell | MySpell

[Jose Da Silva]

On July 18, 2005 01:35 pm, Arsen Kostenko wrote:
> As you might see, I am a newbie here. It eventually happened so, I
> was accepted into Google SoC program
> (http://code.google.com/summerofcode.html,
> http://udu.wiki.ubuntu.com/GoogleSocApplications ). My project is
> SpellChecking bounty for Ubuntu Linux (
> http://udu.wiki.ubuntu.com/SpellChecking ).

Congratulations, and glad to see you will be focusing on Aspell.

> The task is to make a 
> single spell checking system instead of three (Ispell, Aspell and
> MySpell). After some negotiations with a mentor of mine, Aspell was
> chosen as the core for the future single spell checking system.

Your mentor gave you good advice.
...but of course, this is a "biased" [aspell-devel] opinion.   ;-)

> Since I am quite new to Aspell as well, I would be glad if Aspell
> gurus helped me with advice.

I'm not a guru, however, if you are looking to help Aspell and you have 
a solid chunk of devoted time in which you can do some serious amounts 
of improvements (or damage), then this is worth reading:
http://aspell.sourceforge.net/
On that web page, check the "Quick links" to 0.60.3, Manual, Devel Docs. 
Check out the ToDos, Things that need doing, Things I would like...

In terms of the CVS, you may want to note there is a 0.60.3 branch 
(which is public & for distribution) based on 0.60.0 and there is also 
a 0.61. MAIN branch (not quite ready for distribution yet) so some 
things are still maintained in parallel, but some things are also 
diverging due to progress.

> Probably the most challenging part is developing MySpell
> compatibility interface in Aspell.

Due to your interest in MySpell, you will want to view 0.61 (MAIN) in 
relation to 0.60.3 in CVS:
http://savannah.gnu.org/cgi-bin/viewcvs/aspell/aspell/?only_with_tag=MAIN
http://savannah.gnu.org/cgi-bin/viewcvs/aspell/aspell/?only_with_tag=rel_0_60_3

As a graduate, you may not have experience here, so hopefully this 
helps.... There is a lot of code within Aspell, and therefore easy for 
developer A to work on one thing while developer B is working on 
something else, which puts some reasonable distance between coders so 
they don't trip over each other trying to do something.
Maybe your biggest challenge may be trying to work within the tight area 
of MySpell without tripping over someone else's attempt at code unless 
you both think along the same lines.
When coding, in general, it helps if you put together patches in a clear 
concise way which makes it easy to understand/accept.
Tiny 1 or 2 line fixes may be easily understood and accepted, but large 
swaths need time to digest, therefore expect it to take time to accept 
if it gets accepted at all.... sometimes all you need is a small patch, 
sometimes you just have to throw it out and put a new approach.
You have to assume that readers don't know what is in your head when you 
thought of a patch, so, explaining what you do, or why you wish a patch 
to get accepted will help get your patch(s) get through quicker. If 
there are certain sections which may not appear obvious, it helps to 
explain it versus assuming it is obvious enough to understand.
Due to complexity, Aspell lacks automated tests like those found for 
testing glibc functions, but if you have tests to prove code works, it 
makes it easier to verify. Tests could probably be programs, swaths of 
test-text, steps 1,2,3 to follow.... basically anything that makes the 
maintainer's life easier when dealing with what you contribute.

> By this time it would be helpful to learn whether task of this
> kind has ever been carried by Aspell developers.

Hopefully the above pointers help you get started in the right direction 
and you can bootstrap yourself into the right mode of thought.
In terms of the word developerS, it would be nice to have an army of 
developers improving Aspell, but Aspell does not quite fit into the 
popularity league of the kernel, so you may find that it is one 
developer, Kevin, ...... and then, several of us doing the occasional 
contribution(s) and improvement(s) .... this usually means resources to 
answer everyone's questions are often a little thin.
Bootstrapping yourself helps, and hopefully, if you are willing to help, 
then that makes 2 developers ;-)

> In case the answer is 'yes', then whom should I address? 
> In case the answer is 'no', then are there any
> objections/considerations on this idea?

You may assume there is intentions to include some sort of MySpell 
compatibility if you look at 0.61 in relation to 0.60.3.
 Sorry, do not have details to help you in terms of status/progress of 
0.61 in relation to MySpell to complete your mission.