Date: Thu, 29 Jul 2004 17:52:18 -0700 From: David Schultz <das@FreeBSD.ORG> To: Scott Long <scottl@samsco.org> Cc: cvs-all@FreeBSD.ORG Subject: Re: cvs commit: src/lib/msun/src e_scalb.c e_scalbf.c Message-ID: <20040730005218.GA1247@VARK.homeunix.com> In-Reply-To: <4109853F.3090508@samsco.org> References: <200407091001.i69A1A6V051028@repoman.freebsd.org> <41097469.10807@samsco.org> <20040729222415.GA728@VARK.homeunix.com> <4109853F.3090508@samsco.org>
next in thread | previous in thread | raw e-mail | index | archive | help
On Thu, Jul 29, 2004, Scott Long wrote:
> Is this going to hose people doing source upgrades from 5.2.1 to 5.3
> and/or 4.x to 5.3. I know that the latter is going to be tough for a
> lot of other reasons, but loosing the former would be bad. Also, does
> it affect binary compat of 4.x/5.2 programs on 5.3?
Yep, I know, and I intend to fix the problem by having an
implementation of isnan*() in libc and libm. (It needs to be in
each, for different reasons.) The commit you responded to is
actually related to fixing some slightly different source/binary
compat problems; it was another commit that caused the breakage.
The trouble is that there's a lot of cruft we need to support
because we never bump libm's version number. We have to be able
to link *any* of libc.so.{2,3,4,5} with libm.so.2, we need to
support programs compiled with the 4.x math.h running with the
5.X libm.so.2, etc.
As I said before, I'll try to look into it tonight. But at the
moment, I have a lot going on in my day job...
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?20040730005218.GA1247>