On Sat, 2004-10-23 at 12:22, Michael 'Mickey' Lauer wrote:
> Am Freitag, den 22.10.2004, 23:14 +0100 schrieb Peter Flynn:
> > There does seem to be a pattern here in a number of facilities
> > which were working perfectly in 3.2 but are broken in 3.5.1.
> > The difficulty seems always to be in getting enough people (end
> > users) to test new releases, because it means wiping their Z
> > and reflashing, and many people can't do that (including me)
> > because they have come to depend on it for work.
> Yeah, that's the vicious circle.
We could break this if new versions of programs were statically
compiled, allowing existing users to download them and run them,
and find problems with them, without having to wipe and reflash.
Is that technically possible?
> And just for the records, I stated more
> than once that OZ 3.5.1 is more or less created _from scratch_, i.e.
> it's in no way based on 3.2 or 3.3.6-pre1
> (well with the exception of that ancient kernel), which means that it is
> totally expected that things that worked in earlier releases no longer
> work in 3.5.1.
My apologies, I must have missed this while I was offline. That
would explain some of the more surprising problems I have raised.
> If we had more people joining us in the last 12 months, 3.5.1 would have
> been a more polished release. But alas, this didn't happen. And I really
> no longer bother to do much work in acquiring new people. We release
> early and often and if no people send in fixes, the same bugs won't get
> fixed - possibly forever.
What about the position where bugs are at least spotted and recorded
by the user-who-is-not-a-programmer. Is that useful, or is it really
only code fixes that are needed?