| Re: [modeling-pmc] Infrastructure Changes |
Nick,
If the purpose of this exercise is to get the build infrastructure to work against the new structure, I'd suggest taking a snapshot of what's in CVS (as you've done), restructuring it based on what we think the new layout will be, and updating the build scripts, etc. to work against it. We should continue to develop and build against the current CVS structure on dev.eclipse.org, and when the time comes, simply migrate the releng plug-ins from your sandbox and reorganize the project folders on dev.eclipse.org...
Cheers,
Kenn Hussey
Eclipse UML2 Project Lead
Rational Software, IBM Software Group
770 Palladium Drive
Kanata, Ontario, K2V 1C8
T: (613) 599-3980 F: (613) 599-3912
Nick Boldt/Toronto/IBM 07/25/2006 04:05 PMTo
Kenneth Hussey/Ottawa/IBM@IBMCA
cc
Max Feldman <Max.Feldman@xxxxxxxxxxx>, PMC members mailing list <modeling-pmc@xxxxxxxxxxx>
Subject
Re: [modeling-pmc] Infrastructure Changes
Well, as I see it, we have two options...
a) start using the new cvs
b) keep using the old cvs
if (a) - there'd be no need to synch back, but you'd also not get your changes built until after the releng changes were working
if (b) - we'd have to merge the changes over into the new cvs once the releng was ready, but you'd be able to publish builds as we go using the old builder on emf.torolab, and the current cvs on dev.eclipse
I suppose it depends on:
* how urgent the changes are - that is, how urgent they need to be published in new builds,
* if you want them to be part of MDT or part of the current projects then later MDT as well, and
* your comfort level with assisting me in the synch back step if we choose (b).
The other thing to consider is that I'm only looking at moving files around and working on the releng piece. I'm not planning to refactor anything (should I be?) to a new namespace like org.eclipse.mdt.whatnot, so I shouldn't be changing anything in the actual plugin code, just the folders in which it resides and the releng stuff to support those moves.
Once UML2, XSD, OCL and EODM are repositioned in /cvsroot/modeling/mdt/releng and /cvsroot/modeling/commons/releng (or something like that), and we have working builds again, THEN we can perform the code synch/merge (either a straight copy, an rsync, or some manual, labour-intesive diff thing (?)), test out the builds still work, and launch into any refactoring still to be done.
Does that sound like a reasonable approach? Which scenario, (a) or (b) - or another as yet unknown idea, (c) - sounds like the best solution for you?
Cheers,