Re: Introspector & Treecc (was: [DotGNU]Licence question about GNU and GCC)

[James Michael DuPont]

Norbert,

Thank you for your encourgement! 

With the DOTGNU project members, hopefully the
intropspector tool will find an application that will
be a benefit to the entire GNU community. 

> > Even more, Creating bi-directional gateway on the
ast level by interfacing the treenodes from gcc into a
treecc node structure. 
> 
> I like this idea very much... this would create lots
of interesting possibilities.
 
> > Even more, adding in XML dumping and Perl
> interfaces
> > on the tree level to both languages with a
> database
> > representation for debugging!
> 
> Very cool.  Besides making it easier to hunt 
> bugs in the parsing and lexical analysis stages, >
with some good graphical visualization tools, 
> this could empower
> hackers to quickly check whether the current 
> optimizer does a good job on the 
> performance-critical parts of their code.
> If enough hackers do that, we may soon afterwards
> see very good progress on the optimizers front.

The ast-optimisation branch of the gcc is doing that
by de-parsing the code. This is very cool.

An xml representation and a database representation
however allows for style sheets to be applied to use
Rapid application development of prototypes of
optimisation techniques to be built without the
overhead of learning the full complexity of the
system.

> I hereby withdraw my initial skepticism of your
> plan.  I am now convinced that the strategic
benefits for Free Software of having such tools
available are so
> significant that they definately outweigh the
strategic risks that we have discussed.

Thank you very much! That is the most positive thing
that I have heard in a long time. 

We should do a node type by node type comparison of
the tree nodes that you and the gcc use as the first
step of the definition of the interface.

The treecc tool is very cool, and my focus will be on
identifying the areas where we can implement such a
transformation of the trees.

XSLT is such a tool for the XML Document trees,
on an abstract level, the AST tree transformations are
similar to XML tree transformations. 

Maybe a simple set of tranlation rules would be the
ideal representation of the knowlege of the
differences between the two language representations? 

These rules could then be optimised down into
executable code as we go along. But of course this
needs more thinking over.

Best Regards,

James Michael DuPont

=====
James Michael DuPont

__________________________________________________
Do You Yahoo!?
Try FREE Yahoo! Mail - the world's greatest free email!
http://mail.yahoo.com/