RE: Re: [Opencxx-users] refining the Synopsis AP Is

> From: Vladimir Prus [mailto:[email protected]]
> Sent: November 1, 2004 10:30

> On Monday 01 November 2004 17:44, Stefan Seefeld wrote:
> > > From: SF Markus Elfring [mailto:[email protected]]
> > > Sent: November 1, 2004 08:44
> > >
> > > I guess that you prefer the approach that is described by the
> > > papers on the page 
> "http://keystone.sourceforge.net/research.shtml".
> > > Would you like to reuse the software that was developed by
> > > the other project?
> >
> > You asked this question already in another thread. My answer
> > is still 'no'. For reasons see earlier discussions.
> 
> Stefan, do you have a pointer to that other thread? I'm not 
> saying we should 
> be using keystone, or parts of it (have no opinion), but is 
> interested what 
> you think.

Unfortunately it's hard to navigate through the archive of the
discussions, so let me restate my main points here:

* keystone is licensed under GPL, and thus not suitable for a 
  framework that should be usable in proprietary settings, too.

* keystone creates its syntax tree by copying the tokens from
  the input stream (the source code), while synopsis builds
  a parse tree as a structure *on top* of the underlaying buffer,
  i.e. keystone is lossy as far as low level information is concerned.

* as Gilles pointed out, keystone uses two SymbolTable structures,
  one part of the AST, the other part of the ASG. While that may
  be a sensible choice in the context of the keystone parser, I
  don't think that's required for synopsis.

Please note that I'm not familiar with keystone *in detail*, so take
the above with a grain of salt.

Regards,
		Stefan