Return to BSD News archive
Path: sserve!newshost.anu.edu.au!harbinger.cc.monash.edu.au!bunyip.cc.uq.oz.au!munnari.oz.au!quagga.ru.ac.za!Braae!g89r4222
From: csgr@cs.ru.ac.za (Geoff Rehmet)
Newsgroups: comp.os.386bsd.development
Subject: Re: shlib_minor from 0 to 1
Date: 26 Jul 1994 16:54:47 GMT
Organization: Rhodes University Computing Services
Lines: 24
Message-ID: <313f4n$mq9@quagga.ru.ac.za>
References: <3087d6$abn@quagga.ru.ac.za> <310fa1$c76@cleese.apana.org.au> <3112dm$gkr@quagga.ru.ac.za> <CtJoB6.Iqq@cogsci.ed.ac.uk>
Reply-To: csgr@cs.ru.ac.za
NNTP-Posting-Host: braae.ru.ac.za
X-Newsreader: NN version 6.5.0 #4 (NOV)
In <CtJoB6.Iqq@cogsci.ed.ac.uk> richard@cogsci.ed.ac.uk (Richard Tobin) writes:
>In article <3112dm$gkr@quagga.ru.ac.za> csgr@cs.ru.ac.za writes:
>>When I said that 1.1.5.1 binaries would not run on 1.1 was that a 1.1
>>system with lib*.so.1.0 would not be able to run 1.1.5 binaries, which
>>would require lib*.so.1.1.
>Under SunOS, this would only result in a warning. Given that the
>major version is meant to change if the specification of the functions
>change, this seems like a reasonable solution. It's possible that
>something won't run because new functions have been added, but if
>it does run it should run as well as a binary compiled with the old
>library. Why not allow it?
Yes, that is something I would like to look at doing. As every "good"
developer says - hopefully in the next release. (But first I gotta
write a damned thesis ;-)
Geoff.
--
Geoff Rehmet, Computer Science Department, | ____ _ o /\
Rhodes University, South Africa |___ _-\_<, / /\/\
FreeBSD core team | (*)/'(*) /\/ / \ \
csgr@cs.ru.ac.za, csgr@freefall.cdrom.com, geoff@neptune.ru.ac.za