[Date Prev][Date Next] [Thread Prev][Thread Next] [Date Index] [Thread Index]

Re: New libc6 breaks exim?



On Wed, Sep 27, 2000 at 08:46:39PM -0600 , Jason Gunthorpe wrote:
> On 27 Sep 2000, Greg Stark wrote:
> 
> > I can't believe after fucking up the glibc 2.1 release so badly that they're
> > about to do it again with much less excuse. At least with libc itself you have
> 
> I think technically the problem is not glibc, but with our libdb
> package(s). They need to be compiled in a manner that is compatible with

I said that all the time (or at least wanted to express this)

> the packages from potato, this means they need versioned symbols - which
> were probably only turned on as a side effect of the build process..

as part of the glibc, yes

> Furthermore, I think we must follow whatever RedHat has done here, or
> bully them into adding versioning.. If our libdb.* libraries do not have
> the same symbols as theirs do then we have lost inter-dist binary
> compatibility. 
> 
> Someone grab a RH7 and run 'readelf -s' on the libdb.so.2 and .so.3
> libraries and see what there is to see. From potato it looks like this:
> 
>     41: 00005f10   181 FUNC    WEAK   DEFAULT   12 dbopen@@GLIBC_2.0

yes. they have version symbols on their db2 package. But it's created from
glibc-2.1.3 db2 libraray and not the sleepycat upstream, from what I've seen.
And we should probably do the same, or we're looking bad.

				Petr Cech
-- 
Debian GNU/Linux maintainer - www.debian.{org,cz}
           cech@atrey.karlin.mff.cuni.cz

Obviously the only rational solution to your problem is suicide.



Reply to: