|Catastrophic compiler errors email@example.com (Mark Lacey \[MSFT\]) (2002-04-06)|
|Re: Catastrophic compiler errors firstname.lastname@example.org (2002-04-07)|
|Re: Catastrophic compiler errors email@example.com (2002-04-10)|
|Re: Catastrophic compiler errors firstname.lastname@example.org (Andre Vergison) (2002-04-10)|
|Re: Catastrophic compiler errors email@example.com (Shankar Unni) (2002-04-13)|
|Re: Catastrophic compiler errors firstname.lastname@example.org (Joachim Durchholz) (2002-04-16)|
|From:||Joachim Durchholz <email@example.com>|
|Date:||16 Apr 2002 00:51:46 -0400|
|References:||02-04-035 02-04-058 02-04-062 02-04-087|
|Posted-Date:||16 Apr 2002 00:51:46 EDT|
Shankar Unni wrote:
> After all, is there really a fundamental difference between code not
> working because of the writer's mistake, or the compiler's mistake?
I can test my own code by looking at the decision points and setting up
tests of things that I know are likely to break.
I cannot do this for compilers. I have to do "gunshot testing": try a
few things and see whether things start to break. Not very reliable, but
still the best that can be done. (Changing the compiler, or even just
the compiler options, and releasing the result untested would be
inviting disaster, of course.)
Return to the
Search the comp.compilers archives again.