Re: Justifying Optimization

MICHAEL DILLON <mike.dillon@lmco.com> wrote in message news:03-01-088...
> [I didn't know that anyone was still having this argument. While it's
> true that optimizers sometimes have bugs, it's far more common that
> they reveal bugs in application code that does stuff that the language
> spec forbids,


I often thought that it would be extremely valuable to add a command
line option to a complier called "-weird" or "/wierd". What the
option does is makes legal but really weird choices as to how to
implement stuff. For example, in C, you might choose the order of
evaluation of the parameters using a random number generator
(hopefully, each compile generating a different order), or perhaps
have it select weird sizes for integers (if such is allowable in the
language.) Other possiblities might be to randomly select the order of
members in a record, and so forth. You get the idea.


Microsoft goes some way toward that, in their standard debug
configuration they do various weird stuff, such as initializing local
variables to non zero values, or deliberately blasting the data in
free-ed memory.

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