Compiler Bugs, please!

Hi, everybody,


    Yes, I know that most of You are busy making compilers better (or
making better compilers :-). And we all can do with less bugs, be it
in compilers or in any other piece of software, BUT:


    I am interested in learning about compiler bugs. The 'My compiler
crashes when trying to do [...] feature' kind of thing is probably
known to everybody; but what about the more obscure kind, like the
FORTRAN (or BASIC) compilers which allowed one to assign a value to a
constant (check out the RISKS Digest 12.39, Subj. 2.1)? Are there
more incidents like this? A compiler producing a valid binary which
shows some unexpected behaviour when executed, because of a
misconception on the side of the compiler (or its writer :-) on the
language spec, or on the target architecture? What happened when a
portable compiler _was_ actually ported to compile to a new platform?
What about optimisations producing 'interesting' results?


NB, my most recent addition to my collection is at:
http://www.heise.de/newsticker/data/hes-11.11.01-000/


    Huh, many questions. May I ask that You mail contributions to me
directly, so that I can create a summary?


TIA, Yours,
-- Florian
=======================================================================
Florian Liekweg | RISC: (abbr.) "Reject Important Stuff
IPD Universität Karlsruhe | into Compiler"; see also: CISC

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