Path: utzoo!yunexus!ists!helios.physics.utoronto.ca!news-server.csri.toronto.edu!cs.utexas.edu!swrinde!zaphod.mps.ohio-state.edu!think!snorkelwacker!spdcc!ima!esegue!compilers-sender From: preston@rice.edu (Preston Briggs) Newsgroups: comp.compilers Subject: Aliasing Keywords: code, C, Fortran, optimize Message-ID: <1990May29.161200.1900@esegue.segue.boston.ma.us> Date: 29 May 90 16:12:00 GMT Article-I.D.: esegue.1990May29.161200.1900 References: <4886@daffy.cs.wisc.edu> Sender: compilers-sender@esegue.segue.boston.ma.us Reply-To: Preston Briggs Organization: Rice University, Houston Lines: 36 Approved: compilers@esegue.segue.boston.ma.us In-Reply-To: <54866@microsoft.UUCP> In article <54866@microsoft.UUCP> Mark Roberts write: >Actually, it is possible to do a pretty decent job of alias analysis; it is >just very computationaly expensive. There's at least 3 kinds of aliasing I can think of: 1) Aliasing caused by reference parameters (two parameters aliased or a parameter and a global aliased). This is an interprocedural problem and is addressed by Fortran's anti-aliasing restrictions. It's also the problem addressed by the Cooper-Kennedy work. 2) Aliasing caused by arrays (e.g., do A[i] and A[j] refer to the same location?). This is the problem addressed by dependence analysis. Reference Kuck, Bannerjee, Wolfe, Allen, Kennedy, and many others. 3) Aliasing caused by pointers (do *p and q->key refer to the same location?). This is the one that doesn't occur in Fortran. Reference Larus and Hilfinger, Sigplan '88 for example. Also, Chase, Wegman, and Zadeck have a paper appearing in Sigplan '90. >[Now Preston can give us his 'why I like Fortran' spiel ;-) ] How embarrassing. For the record, I dislike Fortran even more than C. Too many people assume because I slam C's pointers that I must therefore like Fortran. The language spectrum isn't so restricted! Support your local alternative language -- SML, Oberon, Clu, Beta, ... -- Preston Briggs looking for the great leap forward preston@titan.rice.edu -- Send compilers articles to compilers@esegue.segue.boston.ma.us {spdcc | ima | lotus}!esegue. Meta-mail to compilers-request@esegue. Please send responses to the author of the message, not the poster.