Subject: Re: i386 elf and libc
To: None <firstname.lastname@example.org>
From: Michael Richardson <email@example.com>
Date: 12/10/1999 13:22:16
>>>>> "matthew" == matthew green <firstname.lastname@example.org> writes:
matthew> probably would have saved some hassle for this, yeah. but there are still
matthew> about 4 architectures (at least) to convert to ELF. besides this transition
matthew> period, there is *every* reason to keep the shared library version numbers
matthew> the same across all ports.
matthew> Okay, how about minor versions? Some of them, libm for instance and bfd are
matthew> identical. I do *NOT* want /emul/aout if I can avoid it.
matthew> i don't understand what you are saying here.
If there were no filename comflicts, then aout and elf libraries could
co-exist in /usr/lib.
matthew> the argument for not changing was that the were no *interface* changes
matthew> in the library. the shared library version is 100% independant of the
matthew> format and 100% dependant on the interfaces present.
matthew> So, is the reason for the renames to have remained in the elf versions is
matthew> that we wouldn't want to maintain two sets of includes?
matthew> the renames remained because the featureset was not changed.
matthew> i agree that in an simple (eg, freebsd single architecture) world revving
matthew> every major library version would avoid the problem. but that's not what
matthew> we have....
Well, I think we should think strongly about bumping major versions for the
next release, and getting rid of all renames. The new libraries are not
compatible with the old ones, regardless of whether the "featureset" has
] Out and about in Ottawa. hmmm... beer. | firewalls [
] Michael Richardson, Sandelman Software Works, Ottawa, ON |net architect[
] email@example.com http://www.sandelman.ottawa.on.ca/ |device driver[
] panic("Just another NetBSD/notebook using, kernel hacking, security guy"); [