On 03/08/07, Rafael Schloming <[EMAIL PROTECTED]> wrote: > Martin Ritchie wrote: > > On 01/08/07, Rafael Schloming <[EMAIL PROTECTED]> wrote: > >> Martin Ritchie wrote: > >>> Can I ask why my commit revision 561578 was rollback? > >>> > >>> If there was a problem with my commit I would expect there to be a > >>> discussion on the list before over 2 hours of my work is thrown away! > >> Your work wasn't thrown away. You can recover it in your local checkout > >> with the following command, get it to build, and then resubmit the > >> changes. Nothing you have done was lost. > >> > >> svn merge -r561364:561365 qpid/java > >> > >>> Rafael as you rolled back the changes can you please explain. > >> The changes didn't build, as I said before there were conflict markers, > >> conflicting imports, calls to nonexistent methods. I did make an effort > >> to fix the problems, but after digging through a number of issues I > >> realized that the code hadn't been completely resolved, or compiled. At > >> that point I realized I had no way of knowing whether it would take 2 > >> hours or 2 days to get the trunk building again, and given how many > >> other people are working on the trunk I really had no option but to > >> rollback the change as I couldn't leave the trunk in that state for an > >> indeterminate period. > >> > >>> I can guarantee that some of the files that you have removed from > >>> trunk will *NEVER* impact the build process. > >> That may well be true, however I thought it better to just rollback the > >> whole change underneath the java dir so that you could easily recover it > >> with the svn command I supplied above and then pick up where you left > >> off in your conflict resolution. > >> > >>> If my changes caused problems with the build then I apologise however > >>> I was unable to fully test the changes because the maven build system > >>> was already broken on trunk. > >> The fact that the build didn't work correctly out of the box on cygwin > >> was my fault. A situation for which I apologize, and have since > >> attempted to rectify. The code, however, was not broken at that point > >> since it was being regularly built on non windows boxes. > >> > >> If the build is broken, please make every attempt to fix it or email the > >> list so that someone else can fix it. Submitting changes on top of an > >> already broken trunk without fixing it first only makes it harder to > >> unravel the reason that the build is failing, and the odds that you will > >> get your merge correct without ever building it are vanishingly small. > >> > >> In the unlikely event that someone needs to break the build with a > >> submit, please let the list know so that the submit can be coordinated > >> with other people's oustanding changes, and a plan can be made to bring > >> the trunk back to working order in a reasonable timeframe. > >> > >> --Rafael > >> > >>> On 31/07/07, Rafael Schloming <[EMAIL PROTECTED]> wrote: > >>>> Martin Ritchie wrote: > >>>>> Hi, > >>>>> > >>>>> I've merged all the changes from M2 to trunk. The merge was mostly > >>>>> painless. > >>>>> > >>>>> The only problems were: > >>>>> > >>>>> - I mvn doesn't seem to build on trunk. > >>>>> - I've not used trunk in a while but it fails to compile the Common > >>>>> package with this error: > >>>>> ( I put error at the end as it is a big stack trace ) > >>>> It looks like this error is due to a problem invoking the python script > >>>> that generates the code for the 0-10 work. The invocation from the pom > >>>> file probably needs to be tweaked to work correctly on windows. > >>>> > >>>> --Rafael > > > > Hi Rafael, > > > > Sorry if I seemed a bit off this morning must have gotten out of the > > wrong side of bed. > > Either that or my windows box was killing me again. > > Anyway no excuse. > > > > I'm sorry I committed the code with so may errors. I did carefully go > > through the diffs between my M2 and trunk after the merge. Most of the > > changes were to do with the exception method changes. > > > > I tried the command (svn merge) from above but it didn't appear to do > > the trick. Once I've cleared my work load I shall take another look at > > doing the merges. I don't think there is much different... perhaps > > just using kdiff and hand merging M2 on to Trunk might be the easier > > approach. > > > > Thanks for attempting to look at the build problems. I'm surprised > > there were conflict marks as both svn and kdiff never spotted them. > > But hey they are just tools :) I'm sure javac would have found the > > problem. I was just too impatient to commit as it kept failing because > > the repository was changed whilst I was doing the commit. Should have > > just waited till the morning rather than trying to commit last thing. > > No problem, I've been there myself. ;) > > I tried out the merge command and you're right, it doesn't get the paths > right the way I specified it. If you do the following it should work > correctly: > > cd qpid/java > svn merge -r561364:561365 . > > The original form I gave you tries to apply the delta to the wrong level > of the tree. If this version doesn't work for you, please let me know. > > --Rafael
Thanks Rafael, I give that a whir shortly. Hopefully we can resolve all these niggly build problems soon. Cheers -- Martin Ritchie
