Re: compiler bugs

"Christopher Glaeser" <cdg@nullstone.com> writes:
>> An optimizer that breaks a program is a bad idea. There are
>> apologists (of program-breaking optimizers) that claim that the
>> program was already broken without the optimizer, because it does not
>> conform to some language standard. But actually the program does
>> conform with the language as it is implemented by the compiler without
>> optimization and it behaves as intended by the programmer, so it is
>> correct.
>
>Can you please provide us with the list of optimizations that will not
>affect the behavior of any program that uses language features that are
>defined by the language standard to have undefined behavior?


*The* list, for all languages? Please contact me for my consultancy
rates:-). Actually, you could make identifying such program-breaking
behaviour an additional service of your business.


Anyway, to give some examples: many forms of constant folding, copy
propagation and common subexpression elimination; but typically a
particular optimization can be performed in a behaviour-preserving way
or in a program-breaking way. E.g., evaluating a constant expression
may result in an exception (e.g., consider "1/0"); proper constant
folding has to preserve that visible behaviour (giving a warning about
the expression may also be a good idea) instead of compiling just
anything because the language standard does not cover this case.


- anton
--
M. Anton Ertl
anton@mips.complang.tuwien.ac.at
http://www.complang.tuwien.ac.at/anton/

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