Re: Was: Debug optimized code; Now: How many compiler bugs have you seen?

On Wed, 24 Sep 2008 12:46:15 -0400, Chris F Clark
<cfc@shell01.TheWorld.com> wrote:


>George Neuner <gneuner2@comcast.net> writes:
>
>> Not to mention that the compiler itself is more likely to have
>> strangely interacting bugs in its higher optimization levels.
>
>People often mention this and having workied on compilers I've seen my
>shared of bugs (from the inside). However, as a compiler customer I
>can only recall two times in my career where I've actually encountered
>a bug that was in the compiler and not in "my" code (the code I was
>working on).


I vaguely remember a case, in the late 90s, using a C compiler for an
80C196KC where the compiler was the prime suspect, but I don't recall
the details or even if it was ever confirmed.


I remember a case where the compiler was blamed, but the real cause
was order-of-evaluation in an expression something like...


    x = fn (lookup [a++], lookup [a++]);


I remember lots and lots of bugs in an interpreted scripting language
for generating reports - that I was writing at the time :-(


Even then, the bugs were in the reporting logic, not really the
scripting language.


Other than that, I think I've had a few bugs where I couldn't build
the code at all because symbols weren't included in the object files
or similar - could have been DLL related. I think that was in Borland
C++ 5-ish (5.01 or 5.02).


Basically, I've never had a confirmed compiler-bug cause a run-time
error in release code that I remember. A few suspicions, heard a few
stories, but nothing ever confirmed. But I have had plenty of
pointer-handling bugs, undefined-order-of-execution issues, and
memset-to-null wiping the virtual pointer issues.

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