Re: [tree-ssa, LNO] Analysis of scalar evolutions and data dependences

[Pop Sébastian <pop at gauvain dot u-strasbg dot fr>]

On Tue, Dec 16, 2003 at 06:06:31PM -0500, Diego Novillo wrote:
> On Tue, 2003-12-09 at 04:59, Pop Sébastian wrote:
> 
> Very nice.  I would strongly encourage you to create a sub-branch.  The
> patch is massive and it is very difficult to get an idea of what it does
> by just staring at it.
> 

Okay, I will do this.  What will be the name of the new sub-branch?  I
have proposed loop-nest-optimizer, is it ok?

> tree-chrec.c could use a bit more documentation.  Several functions have
> no documentation at all.  The routines seem mostly self-explanatory, but
> still.
> 

No problemo.  I didn't noticed it.  I have also detected some places
where the comments were outdated, I will correct these too.

> The high-level overview in tree-data-ref.c and tree-scalar-evolution.c
> is helpful.  Thanks.  Any reason you chose Allen and Kennedy's book?  I
> haven't read it.  Is it worth getting?
> 

There is no particular reason for choosing the Allen and Kennedy's
book.  I put a reference on it because I was reading it when I wrote
the code, but I think that it would be better to have also freely
available pointers to articles on citeseer.  I will add some more
references.  

> What's the idea behind tree-fold.[ch]?  Are you trying to replace
> fold-const.c or are these functions specific for data dependency?
> They look like generic folders to me.
> 

Yes, they are generic folders: it's just an interface that has been
rejected? some time ago.  If you want I can remove it and use the
standard "fold (build (PLUS_EXPR, ...))".  I still think that we need
a better interface to the folder.