> On Wed, 16 Jul 1997, Harald Koenig wrote:
>
> > On Jul 16, Alan Cox wrote:
> >
> > > 10 people. Some of the bugs that show up in 2.0.x show up in configurations
> > > that maybe 1 in 1000 or 1 in 10,000 users have and perhaps only after a
> > > week of continual load. On a 10 user sample 2.0.30 is probably rock solid
> > >
> > > A lot of people don't seem to realise just how tricky it is building a
> > > stable across all platforms/configurations system.
> >
> > well spoken. and I'd like to add: ... how tricky it might be
> > to fix bugs which one can't reproduce himself!
>
> Hi. Reading this flame, I couldn't resist throwing in my 2 cents. I hate
> to do this, because it is offtopic, but here goes.
this was not a flame at all (usually I mark them in public mail, just to be sure!).
reading my 2 lines over and over again I still don't know if you think I flamed Alan,
all developers, or all non-developers on this list ?!
can you please give me a hint why *you* think this is or should be a flame ?
> First, I run a server, running 2.0.30, with a usual load of maybe 5-6
> people.
...
fine for you. but what does this say in this thread's context or what does it proof ?
I know several severe bugs in 2.0.30 but I'm only able to reproduce just one of them,
and that not with "normal" operations. for me, and for our institute at university
2.0.30 is stable (for non-SMP only!). that's not the point here...
> The point of this, is to mix some praise in this flame of anger if you
> will.
again: no flame and no anger! just my statement that it's *very* hard to
fix *known* bugs which can't be reproduced be the developers themself,
even if for some users the machine crashes every minute...
and I think I know what I'm talking about: I've been writing XFree86 Xservers
for S3 cards/chips for ~5+ years now, and there have are *many* real problems
which I never had been able to reproduce myself (e.g. lockups which only
happen for very specific cpu+mainboard+chip+vendor/card combinations)
but which have to be tracked down and fixed anymore.
while I can understand some users saying "hey, this bug is so obvious [for me]"
everyone should keep in mind that developers might have different setups.z
and same is true for kernel things, and even for such easy tasks as compiling
your own kernel (just remember how often you can read questions about
problems linking the kernel just because they use too old binutils --
and there may be more sublte problems like different gcc versions/bugs around)
Harald
--
All SCSI disks will from now on ___ _____
be required to send an email notice 0--,| /OOOOOOO\
24 hours prior to complete hardware failure! <_/ / /OOOOOOOOOOO\
\ \/OOOOOOOOOOOOOOO\
\ OOOOOOOOOOOOOOOOO|//
Harald Koenig, \/\/\/\/\/\/\/\/\/
Inst.f.Theoret.Astrophysik // / \\ \
koenig@tat.physik.uni-tuebingen.de ^^^^^ ^^^^^