Re: [orca-devel] A Couple of Ideas

[Willie Walker <William Walker Sun COM>]

Hi Storm:

Thanks for chiming in!

> I think that after the first initial setup, if Braille isn't selected as 
> a method of output, it shouldn't even be included when Orca starts.  The 
> same would go for magnification.

Agreed.  It goes a bit beyond just importing the modules, though.  What 
we also also in the current code is that the output for braille and 
speech are also computed each time Orca wants to present something.  The 
move to presentation "channels" (if we do them) would be one effective 
way to avoid the extra work: don't load/use channels the user doesn't want.

> My next idea is a plugins directory for Orca.  This directory would 
> replace the current orca-customizations.py file.  This would also be in 
> addition to the current orca-scripts directory.  So, orca scripts could 
> be written as plugins for orca.

Eitan's proposal in http://bugzilla.gnome.org/show_bug.cgi?id=404422 
provides us with a way for scripts to be somewhat automatically detected 
and used.  With this, at least scripts could be automatically detected 
and used if you plopped them in the right spot.  From there, plugins are 
something that could similarly be detected and loaded by scripts, though 
the issues surrounding that need to be worked out.
 > Python interacts rather nicely with C++.  Could some of the code that
> doesn't change much be changed over to c++?  Because it is a compiled 
> language, it would in theory run faster than Python code.

I believe the main issues with performance are related to: 1) repeating 
the same work unnecessarily, and 2) communication between Orca and the 
application.  As such, I don't think the performance issues we see will 
be resolved by C++.  Plus, the complexity C++ will add will probably 
make the code less maintainable and buggier.  :-(

Thanks!

Will