On Tue, 2008-07-01 at 15:24 -0400, michael goulish wrote: > On Tue, 2008-07-01 at 16:31 +0100, Aidan Skinner wrote: > > On Mon, Jun 30, 2008 at 9:31 PM, michael goulish <[EMAIL PROTECTED]> wrote: > > > > > On Mon, 2008-06-30 at 16:18 -0400, michael goulish wrote: > > > > > >> > > >> OK -- attached please find "rhel4_goodies.tar.gz" > > >> It has a README file in it -- but here's the README file: > > >> > > > > > > I just made a JIRA for this and attached the same tar.gz file. > > > It is QPID-1162 . > > > > That's awesome, thanks! Any chance of getting those comitted to trunk? > > I'd rather not have to run the unit tests against a patched build if > > it's avoidable... > > > > - Aidan > > > > Aidan -- > > Sadly, I do not have the Power to Commit. > > > Also. there are a couple things that the patch does that you committers > will want to consider. > > > 1. it removes -Werror from configure.ac, so the boost warnings don't > cause errors. > > > 2. it has some icky options-processing code that I wrote because the > Boost options processing package in 1.32 was buggier than a ten-dollar > motel room. > > Well, maybe not *that* bad, but it had no concept of letting you say > "don't worry about unknown options". That meant that the idea of > loading modules after start-up -- where the modules declare their own > sets of program-options -- didn't work very well. Didn't work at all. > > > > So please consider those facts, and then you committers decide. ( I > would be thrilled. ) > > > ------------------------------- Mick .
We definitely can't disable -Werror in general, although we could do a configure.ac test to disable it on RHEL4 only. The icky code can be split out & conditionally linked/compiled. However the other thing that's in the tarball is a bunch of boost headers from a more recent version than is available on RHEL4 at the moment. I have no clue what the legal ramifications of committing boost headers to an apache repository might be, and no desire to rewrite those bits of boost in Qpid just for RHEL4. If there is a clean way to get all this committed it would be great, building from a patched checkout on RHEL4 is a mighty pain in the ass on my side as well. Any suggestions?
