Re: Questions at my kFreeBSD FOSDEM talk
2011/2/6 Axel Beckert <abe@debian.org>:
> Another question was an old one and despite everyone always tells me
> the same, I still wasn't able to follow the arguments myself and
> therefore can't repeat them properly: What exactly was the big problem
> with the approach of using the BSD libc as done until 2002 in the
> Debian GNU/FreeBSD (no "k") port?
It wasn't a specific problem. I remember Nathan Hawkins worked on
a port to FreeBSD libc. He found that the differences in userland
led to considerable amount of porting work. He made some progress
(I think he even wrote a compat library for UTMP). Then Glibc was
ported to kFreeBSD by Bruno Haible, and Nathan moved on to it
(he'd leave shortly afterwards due to lack of time; before that,
he made the first version of the kfreebsd-8 package; you'll find
his name in debian/changelog).
Another port (to kernel of NetBSD) also used BSD libc. It found
similar problems and eventually stalled.
In short: there was no fundamental blocker that made a port to
BSD libc impossible. My impression is just that different approaches
were tried and this is the one that worked best.
--
Robert Millan
Reply to: