> > > > > For QPID-4417 on on RHEL5 I believe the Swig package will have the > > > > > proper fix backported and released in future. > > > > > > > > I'll be great to get SWIG fixed upstread in RHEL5, however its not > > > > currently fixed, plus some people might not update their packages > > > > anyway, so RHEL5 users will almost certainly get subtly broken > > > > bindings. Wouldn't it be better to reintroduce the minimum version > > > > check - only means when people are building Qpid from source on RHEL5 > > > > they'll also need to build SWIG from source, which they could even keep > > > > in their home directory. > > > > > > It won't matter if someone running RHEL5 hasn't updated their Swig > > > version: Swig is a build dependency, not a runtime dependency, so it > > > only matters that the RHEL5 box on which the package is created has the > > > updated Swig package with the backported fix. > > > > I get that, but currently nobody's RHEL5 has a patch and AFAIK there are no > > official or unofficial Qpid RPMs so most RHEL5 users will compile from > > source and get broken bindings. What are the downsides of re-introducing > > the minimum version check (it'll only affect build, not runtime)? > > If we re-introduce the check then it won't build on RHEL5 since the only > available version of Swig for it is 1.3.29.
It's your call, but as I see it there are two options: 1) Status-quo (0.20 and 0.22) - build fails on RHEL5 with message that SWIG is too old. User can then temporairly build newer SWIG from source in their home directory as a build dependency 2) Allow RHEL5 builds to be subtly broken until newer SWIG package available and assume people who build from source update their packages Cheers, Jimmy --------------------------------------------------------------------- To unsubscribe, e-mail: [email protected] For additional commands, e-mail: [email protected]
