Bug#244897: Broken Debian httpd/APR packaging

[Joe Orton <jorton@redhat.com>]

On Thu, Aug 19, 2004 at 04:00:00PM +1000, Adam Conrad wrote:
> Joe Orton wrote:
> > 
> > 1) change the libapr and libaprutil sonames
> 
> Is there really any guarantee of ABI compatibiltiy between different
> deistributions' use of an SONAME?... Checking around, at least Gentoo
> and Slack seem to compile with LFS as well.
>
> When we compile C++ libraries with a new compiler, we break the ABI, and
> while we change the package names to smooth transitions, we don't
> actually change the SONAME.  I don't see how this is terribly different,
> except that we felt we could get away with not changing the package
> name, because we'd never had that package in a stable release, and we
> can recompile/fix everything in time.

And the libstdc++.so soname changes when the C++ ABI changes.  Not sure
how this is relevant.

> > 2) change the httpd major MMN to be some non-standard 
> > integer, since you 
> > are breaking binary compatibility with upstream and third-party binary
> > modules will crash and burn otherwise.
>
> In light of the above, how can one possibly guarantee that third-party
> binary modules would ever work anyway?  I realise that as a RedHat guy,
> you're likely used to 3rd party binary stuff always working, but that's
> because RedHat is used as a reference system.

In light of what? The fact that there exists some minority distributions
already doing the wrong thing? If you change the ABI you *must* change
the major MMN.  That is what it is *for*.

> Honestly, while I agree with the spirit of what you're saying, the
> practical reality is far from agreeing with it.

Well, you lose support from upstream then (I speak as an upstream
APR/httpd developer).  And you risk triggering all the incompatibilites
between libraries e.g. used by PHP which are compiled with a 32-bit
off_t vs programs using a 64-bit off_t, but presumably you know all this
already.

Regards,

joe