On 20 August 2014 18:17, Fraser Adams <[email protected]> wrote:
> On 20/08/14 17:41, Robbie Gemmell wrote: > >> On 20 August 2014 17:18, Fraser Adams <[email protected]> >> wrote: >> >> I'm assuming that I'm not the only person this is broken for? Clearly >>> someone has been messing with the Java Broker Model stuff again ;-> >>> >>> >>> No, you arent the only person, wasnt working here either prior to your >> change, and Rob has now fessed up I see ;) >> >> I assume it was recent as I have used it myself recently when finishing >> off >> the new packaging. >> > Yeah it was no bigee, I'm trying to reboot my brain after my holiday so > was "quite sad" that it didn't compile, I'm glad it was a simple fix :-) > > > >> There is a CI job to build it to ensure it at least compiles (it could do >> more if there were automated tests), however it looks like it is only >> being >> run when the QMF bits are changed, whereas it should also (and mainly, >> given the respective change frequency) be run whenever the broker is >> changed. >> > Yeah that'd be good. > > >> Look like it was just a change of the Port class on the Model that >>> changed >>> getAvailableProtocols to getProtocols, I've fixed this on trunk under >>> QPID-6024. >>> >>> >>> In future I'd suggest not creating a new JIRA for this kind of >> compilation >> fix, but rather change it on the original JIRA it was broken via if its >> known which one caused it and there isnt some aspect to it that makes it >> seem like an 'unrelated' fix (other things like the original JIRAs being >> in >> a previously released version obviously also being a reason to create a >> new >> JIRA). >> > Fair enough, though TBH finding the original Jira that broke it would > probably have taken me an age - as I say my in-box was swamped after a > couple of weeks away so I mostly just dumped the Jira mails that didn't > *look* relevant :-) > > Hehe :) In some cases it could definitely involve a bit of work, but in the broken compilation case it should be fairly easy hehe, in this case by looking at the version history of the interface that changed, either locally with git or svn, or using one of the repo web interfaces available (viewvc or the github mirror for example). For harder things there is git bisect. > > >> Not having really used it, that seemed like the simplest type of thing to >> try as a smoke test...typical of me to pick one thing that doesnt work :P >> > I'm glad that you did, as I say I'm a bit ashamed that I didn't notice it > myself, it must have always done this with that Java Broker/QMF Plugin. > Interestingly the "canonical" Python qpid-config didn't barf with null > arguments. > > The Java QpidConfig is mainly intended to illustrate how to use QMF with > java (especially the method invocations which are a bit tricky). It's a > little out of step with the python one now and it needs a bit of a refresh > really. > > > >> I really need to do proper mappings from the Java Broker arguments to >>> QMF2 >>> equivalents, but for now I'll populate Queue and Exchange arguments with >>> an >>> empty Map and I'll also add some defensive code to QpidConfig >>> >>> Great >> >> Just fixed this stuff (QPID-6025 <https://issues.apache.org/ > jira/browse/QPID-6025>) on revision 1619161 > > Excellent. I have reviewed it and mentioned on the JIRA I think you should request it for 0.30 :)
