[gnuspeech-contact] Re: cvs

[Steve Nygard]

On Oct 15, 2006, at 11:36 PM, Gregory John Casamento wrote:

> All,
>
> Since we're discussing changes that should be made to the current  
> source,  I would like to suggest that we restructure how the code  
> is built.
>
> A lot of code seems to be shared between the tube model and other  
> parts of GNUspeech in a somewhat non-conventional way using Xcode's  
> ability to include code from any directory (not just the project  
> directory).   I propose, instead, that we examine what the comment  
> portions of the application are and pull them out into a framework  
> that all of the apps can link against.
>
> This is a much more logical structure as it:
> 1) Makes GNUspeech's core code reusable by applications outside of  
> those in GNUspeech
> 2) Is a more conventional way of structuring the code
> 3) Makes it easier to build on GNUstep.

Absolutely, that's the way it should be structured and now is a fine  
time to do it.

It looks like everything in softwareTRM (except TRMSynthesizer and  
main) should be turned into a little library, and then the command  
line softwareTRM program can just link against that.

From Monet, all of the model classes and the things they depend on  
can be moved into a framework, and it'll need to link against the  
library from above.  It probably also needs to pick up TRMSynthesizer. 
[hm].

Most of the stuff from the little GnuSpeech app (GS*Dictionary,  
GSSuffix, TTSParser and dependancies, and probably also the  
dictionary itself) should be moved into a framework.

I'm not sure if we want just one framework (GnuSpeech.framework),  
that includes all of the Monet model code, the dictionary code, and  
whatever else is appropriate, or multiple frameworks...  I think I'd  
lean towards just the one GnuSpeech framework for most of the stuff.

Anyway, I don't really have the time to do this myself, so these are  
just suggestions.

--
Steve