[Date Prev][Date Next] [Thread Prev][Thread Next] [Date Index] [Thread Index]

Re: 2 days till Bug Horizon

On Thu, Mar 02, 2000 at 10:44:16AM +0100, Josip Rodin wrote:
> On Thu, Mar 02, 2000 at 10:20:44AM +0100, Stephane Bortzmeyer wrote:
> > > Almost none of the bug reports against gcc or g++ describing internal errors
> > > are valid, in a sense that they contain a valid and shortest possible test
> > > case, and/or a backtrace, and any other things GCC upstream people would
> > > need to fix the bugs. 
> > 
> > It is not realistic. Almost all "internal errors" of the compiler are
> > triggered only by large and complicated programs. Most of the time, the
> > error disappears when you try to distill the test case.
> I've seen several bug reports describing internal errors, and they included
> kilobyte-long testcase. I guess it's a matter of knowing the language and
> the compiler enough to produce one.
Uhhh... just one note on this... I happen to have an amd k6 233 which gives me
at least on or two 'internal errors' on every kernel compilation. Just looking
what file it was, removing the (obviously corrupt) .o file, and restarting
compilation works.

I've seen this on both slink and potato. A friend of mine with slink on amd k6-2
300 has the exact same problems...



PS. I don't have this on any of my Intel machines (486, pentium, pII).

Reply to: