Re: [RFA] Improved topformflat.l for delta for contrib/ inclusion

[Richard Guenther <richard dot guenther at gmail dot com>]

On 8/26/05, Daniel S. Wilkerson <dsw@cs.berkeley.edu> wrote:
> Scott McPeak wrote:
> 
> > > This is the hacked topformflat.l I use all the time reducing C++
> > > testcases.  It is quite improved over the original one distributed
> > > together with the Delta implementation at http://www.cs.berkeley.edu/~dsw/
> >
> > Thanks!  Next time I minimize something I'll try your topformflat instead,
> > and assuming it's good to go we'll probably include it in the next delta
> > release (who knows when that might be though...).
> >
> > I still have on my future projects list to build a delta tool that really
> > understands syntax, including static semantic dependencies, as such a tool
> > ought to blow away delta+topformflat, but the latter does just well enough
> > that I keep putting it off.  :)
> 
> We do have plans to improve delta.
> 
> > > May I propose to stick this into contrib/ ?  Scott, Daniel - do you have
> > > any objections to this showing up inside gcc?
> >
> > It's all good by me.
> 
> Please explain why you want to put a better top form flat into gcc, but not all
> of delta?  And then if it is properly part of delta, I think I might rather
> that I just shipped it with delta and that you pointed at delta.  Why should it
> go into the gcc distribution?  And if so, perhaps 1) we put the new top form
> flat into delta, and then 2) you just distribute the whole new delta with gcc,
> as an embedded but semantically distinct project?

It was just to have a distribution place for the improved topformflat, so others
could get hands on it.  I'm happy, if you release a newer version of delta
with the improvements, too.  I don't know if it's appropriate to put all of
delta into gcc/contrib, but even to the topformflat inclusion we didn't get
any comment from GCC people yet.

Do you plan to release an updated version of delta in the near future?  Would
it be possible to stick it somewhere more public, like sourceforge or savannah?

I guess for the time being I'll write up a gcc wiki section on
testcase reduction
and point to the mailinglist archives for the improved topformflat.

Thanks,
Richard.