Re: (no subject)

[Mark Slicker <jamess1 wwnet net>]

On Thu, 26 Apr 2001, Marat Boshernitsan wrote:
> 
> Yes, we are listening!
> 
> There are several reasons for holding up the release of Harmonia.  Some are purely pragmatic: lack of documentation and of application showcasing the features of the framework.  Our hallmark application, the harmonia-mode for XEmacs, has been delayed by some internal re-architecting and we are hesitant to release the analysis library without being able to show off at least some of its features.  At this point in time we are hoping to release Harmonia at some point during the summer, at which time at least some of the documentation will also become available. (In particular, I am finishing up my M.S. report describing the framework at high-level.)
> 
> We are also considering going open-source, but at this point we are still debating.  On the one hand, it is true that we could benefit from involving external developers: there's a number of small (and  not so small) engineering projects that would take Harmonia to the level of something actually useful for building IDE-like tools.  One the other hand, we need to regain strong control over the framework (esp. at the design level) in order to adapt it to our research, yet, being a small research group with very limited funding we do not have the resources to manage a real open-source effort.  
>

Perhaps gpf could become the more practical open-source version of
Harmonia. This might allow your project to have see some of the
engineering related aspects being solved while maintaing control of the
research aspects. We don't necessarily need the source to do this, however
I think that would help us out a great deal. Also if we did have the
source initially to work from, the code bases would have a lot more in
common and there would be a greater ability to share code in the future
between projects.
 
> It is also not clear how much the community could benefit from having a framework such as Harmonia available to them.  Keep in mind that many aspects of the framework are incomplete (e.g. the XML format for exchanging program data that was mentioned here is prototyped, but not actually implemented), we do not know whether it is sufficiently scalable, and we are less than convinced that it is usable for building real industrial-strength applications.  
> 

Yes, I understand. I haven't gotten to the point where I can pratically
test gpf, but the main thing that concerns me is space usage. The
versionined parse trees look particularly expensive. I guess that
maybe one advantage of not being a research project (though I am very
interested in the research!), is that we make adaptations based on what is
pratical to implement. 

Mark