dotgnu-general
[Top][All Lists]
Advanced

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: [DotGNU]Working Groups plan v2


From: James Michael DuPont
Subject: Re: [DotGNU]Working Groups plan v2
Date: Thu, 17 Oct 2002 23:22:29 -0700 (PDT)

--- Eric Altendorf <address@hidden> wrote:
> On Wednesday 16 October 2002 14:48, James Michael DuPont wrote:
> > I would like to see an interface repository based on
> > storage of the classes dumped into a graph in the database is
> > queried by a dotgnu webserver. Like the vrs server, but using
> > persistant storage.
> > Kinda like a huge gnutella network where the lists of mp3 float
> > around the net, the class descriptors would be exchanged.
> 
> Could you explain a little bit more about what you have in mind here?

Yes : I have been experimenting with extracting information about
programs from the compilers directly. This produces huge amounts of
data. Varible Declartion, Function Declaration, function bodies,
expression objects. If order to effectivly use this data, you must be
able to seach it. Give that there are so many programs, and such much
data, and so many versions of code. You would need 1000* as much data
as used by the package managment systes like debian. This information
is not all needed at once. 

>  
> What servers would participate?  
An author would serve the code that he writes, or give it to a
preferred server, that would replicate over the net as needed.

>Whose classes would be stored?  
Any classes that are published into the system, I would say the
important parts of all the major GNU tools would be a great start.

>Who 
> would access the class descriptions and why?

You might ask, who is calling this function? What fields are provided
by class X? What classes have a field like Y? what methods access
varible z?  What classes are in package U?

You would access this data from your browser, via webservices, or you
would cache important information into memory, even generate an IL
metainformation database out of it.

You can use more detail for syntax highlighting, function completion in
the editor, basically any tool that needs to access meta data.

This goes back into the meta-data api and development tool api
discussions I have been involved with on this list.

mike


=====
James Michael DuPont
http://introspector.sourceforge.net/

__________________________________________________
Do you Yahoo!?
Faith Hill - Exclusive Performances, Videos & More
http://faith.yahoo.com


reply via email to

[Prev in Thread] Current Thread [Next in Thread]