Re: Intermediate Representation (VBDis)
12 Oct 2001 00:22:01 -0400

          From comp.compilers

Related articles
Intermediate Representation (2001-10-10)
Re: Intermediate Representation (Jon Beniston) (2001-10-12)
Re: Intermediate Representation (2001-10-12)
Intermediate Representation napi@rangkom.MY (1990-08-07)
Re: Intermediate Representation briscoe-duke@CS.YALE.EDU (Duke Briscoe) (1990-08-08)
Re: Intermediate Representation (Preston Briggs) (1990-08-08)
Re: Intermediate Representation (Michael O'Donnell (508)392-2915) (1990-08-09)
Re: Intermediate Representation grover@brahmand.Eng.Sun.COM (1990-08-09)
Re: Intermediate Representation (1990-08-10)
[19 later articles]
| List of all articles for this month |

From: (VBDis)
Newsgroups: comp.compilers
Date: 12 Oct 2001 00:22:01 -0400
Organization: AOL Bertelsmann Online GmbH & Co. KG
References: 01-10-028
Keywords: analysis
Posted-Date: 12 Oct 2001 00:22:01 EDT (Simon) schreibt:

> Is there in fact a difference in the way the two would be
> represented at this stage of the compiling process due to the
> dependance within the loop?

I don't think that there exists any difference in the compiler internal
representation of your loops. Loop optimization is done later, together with
other optimizations. I remember a simple benchmark program, which took from 3
seconds up to over one hour to excute, depending on the optimizations of the
compiler. No compiler found the best solution then, since even the most clever
compiler retained a loop to increment the loop counter.

Back to your question, a compiler must retain the source code structure at
least until all checks have been performed, in order to supply the location of
possible warnings or errors. After this stage any reduction of the syntax tree
is possible.


Post a followup to this message

Return to the comp.compilers page.
Search the comp.compilers archives again.