Path: utzoo!utgpu!news-server.csri.toronto.edu!cs.utexas.edu!yale!cmcl2!lanl!lambda!jlg
From: jlg@lambda.UUCP (Jim Giles)
Newsgroups: comp.lang.misc
Subject: Re: interprocedural alias analyis and pointers
Message-ID: <14326@lambda.UUCP>
Date: 10 Apr 90 21:13:40 GMT
References: <6526@brazos.Rice.edu>
Lines: 24

From article <6526@brazos.Rice.edu>, by preston@titan.rice.edu (Preston Briggs):
> [...]
> Well, we can do better than link- or load-time optimization.
> At Rice U., people (Callahan, Cooper, Kennedy, Torczon, and others)
> have been working on interprocedural analysis for several years.  
> They've discovered fast algorithms to do, among other things, 
> interprocedural alias analysis.  Further, they know how to handle 
> separate compilation (despite the ivory tower that people are 
> beginning to invoke).  Finally, the algorithms are used at Rice
> and in industry (to further spite the anti-intellectual heathen
> among you).

I am, of course, interested in the work.  And I will go immediately to
the library to get copies of their previous papers.  But, I don't see
how you can to interprocedural analysis where separate compilation is
involved.  Separate compilation implies that the compiler is completely
unaware of the content of procedures other than the one currently
being compiled.  In fact, all the procedures in a given program may
have been developed _and_COMPILED_ at different sites.  The only thing
the local user has are the libraries and a loader to put them together.

Anyway, just trying to get the straight story.

J. Giles