|Nondeterministic compilers? email@example.com (glen herrmannsfeldt) (2004-02-26)|
|Re: Nondeterministic compilers? firstname.lastname@example.org (Christian Bau) (2004-02-26)|
|Re: Nondeterministic compilers? email@example.com (2004-02-27)|
|Re: Nondeterministic compilers? firstname.lastname@example.org (Franck Pissotte) (2004-02-27)|
|Re: Nondeterministic compilers? email@example.com (Alex Colvin) (2004-02-27)|
|Re: Nondeterministic compilers? firstname.lastname@example.org (Joachim Durchholz) (2004-03-02)|
|Re: Nondeterministic compilers? email@example.com (Walter) (2004-03-02)|
|Re: Nondeterministic compilers? firstname.lastname@example.org (2004-03-06)|
|Re: Nondeterministic compilers? email@example.com (norm rubin) (2004-03-11)|
|From:||firstname.lastname@example.org (Nick Maclaren)|
|Date:||6 Mar 2004 14:02:51 -0500|
|Organization:||University of Cambridge, England|
|Posted-Date:||06 Mar 2004 14:02:51 EST|
glen herrmannsfeldt <email@example.com> writes:
|> Does anyone know about any compilers that are not deterministic? I
|> could imagine using an algorithm for the optimizer such as simulated
|> annealing that could result in different object code compiling the
|> same source. I know this is done in IC and FPGA layout routing, but I
|> don't know that I have heard of it in high level language compilers.
IBM H Extended (Enhanced, I think) for Fortran was. I reported a
low-priority bug where it generated two different instructions at
one point, depending on whether the routine was first in the file
or not. It didn't get through IBM's support mechanism ....
No, it wasn't deliberate.
Return to the
Search the comp.compilers archives again.