When I've branched Apache Subversion projects, I haven't used a branch manager. Whoever checks in a transaction that needs to go to the branch can take on merging to the branch as well.
IMO, a branch is exactly the right thing, and there's no particular reason to wait. They're cheap, fast, and merging in transactions is easy. Tags should be created off branches, not the trunk. If nothing else, this lets us more aggressively check in 0.9 work that might be destabilizing. -- Adam On Tue, Nov 18, 2008 at 3:14 PM, Ian Boston <[EMAIL PROTECTED]> wrote: > Just a thought, > could we do an RC1, RC2 and final, as a tag, rather than branch, then we > wont have to find a branch manager to manage the branch. > > I think that there were some maven build breaks due to mvn 2.0.6 and > features.txt not being packed. I fixed that this morning (currently on the > west coast US). > > You have to use maven 2.0.8 or later. > > There is one more on the list at the moment, but I am not certain what is > (yet) > > Ian > > On 18 Nov 2008, at 14:32, Chris Chabot wrote: > >> branching sounds good, I don't know if people have any big supprising >> patches scheduled, but if so please hold them back until we did branch :) >> >> I'd say lets branch tomorrow afternoon? If everyone else feels ready for >> that too. >> >> Then we can take a few days to kick the tires and make sure nothing >> squeaks >> to hard or falls appart, but in general I think our code tends to be >> stable >> anyhow so I don't expect to many surprises. >> >> Ps, we did have a couple of 'maven build broke' mails on the shindig-dev >> list in the recent weeks, is there anything in there that should worry us >> before releasing? >> >> On Tue, Nov 18, 2008 at 8:45 PM, Kevin Brown <[EMAIL PROTECTED]> wrote: >> >>> I'm back now. >>> >>> What if we branch (code freeze) now and you patch the serialization into >>> that? >>> >>> We really do need a code freeze at minimum this week. >>> >>> On Tue, Nov 18, 2008 at 7:15 AM, Ian Boston <[EMAIL PROTECTED]> wrote: >>> >>>> Chris, >>>> >>>> I don't think that we (I) have fully resolved the XML serialization >>> >>> issues >>>> >>>> in the Java OS side. >>>> Everything works and is compliant with the 0.8.1 text, but there are >>>> some >>>> ambiguities, and the XSD is certainly not the one in the Spec (which >>> >>> didn't >>>> >>>> match the intent of the text).... and may be different from the PHP >>>> code. >>>> >>>> There is also an open issue of the discussions for this in 0.9 pending >>>> feedback from the PoCo community. >>>> >>>> ---- >>>> We also have 1 blocker bug in the Java Side >>>> https://issues.apache.org/jira/browse/SHINDIG-497 >>>> >>>> This could be invalid, but certainly need checking IMHO. >>>> >>>> We also have a number of blocker feature request, which may have been >>>> categorized incorrectly >>>> https://issues.apache.org/jira/browse/SHINDIG-561 >>>> https://issues.apache.org/jira/browse/SHINDIG-560 >>>> https://issues.apache.org/jira/browse/SHINDIG-559 >>>> >>>> I would be happy to go with these as known issues in the 1.0 release, >>>> although I feel that *I* should really fix/check 497 upto the current >>>> XML >>>> state. >>>> >>>> ----------- >>>> >>>> I can do a maven assembly plugin to generate the artifacts (including >>>> the >>>> php tarball) and check that the release plugin works. >>>> >>>> I assume that the php tarball just need to be the source tree, excluding >>>> the java subtree... or is it more complex than that ? >>>> >>>> Ian >>>> >>>> >>>> >>>> On 18 Nov 2008, at 08:07, Chris Chabot wrote: >>>> >>>> I was just thinking the same ... the php part is pretty much ready for a >>>>> >>>>> release, I need to apply 2 or 3 more patches that are in JIRA and it's >>>>> good >>>>> to go. >>>>> >>>>> On the java side their ready to rock too, however the guy who's the >>>>> most >>>>> likely packager of the release, Kevin Brown, has a family emergency >>>>> last >>>>> and >>>>> this week. >>>>> >>>>> So my suggestion is to wait a few days more to see if he'll be able to >>>>> join >>>>> the release effort, and then set this in motion. >>>>> >>>>> Will be nice to finally have a real 'release' out there :) >>>>> >>>>> -- Chris >>>>> >>>>> On Tue, Nov 18, 2008 at 7:14 AM, Brian McCallister <[EMAIL PROTECTED]> >>>>> wrote: >>>>> >>>>> Nudge, 17th is here and passing :-) >>>>>> >>>>>> On Sun, Nov 2, 2008 at 5:38 AM, Chris Chabot <[EMAIL PROTECTED]> >>> >>> wrote: >>>>>> >>>>>>> So nov 17th sounds good to everyone? (conditional on having no major >>>>>>> glitches in the meantime ofc) >>>>>>> >>>>>>> I think Tarball for PHP and normal maven release for java sounds good >>> >>> to >>>>>>> >>>>>>> everyone too right? >>>>>>> >>>>>>> We should also consider who will write the big announcement post (and >>> >>> no >>>>>>> >>>>>> I'm >>>>>> >>>>>>> not volunteering :)), since it'll be quite a milestone, our first >>> >>> 'real >>>>>>> >>>>>>> incubation' release! >>>>>>> >>>>>>> -- Chris >>>>>>> >>>>>>> On Sun, Nov 2, 2008 at 6:25 AM, Kevin Brown <[EMAIL PROTECTED]> wrote: >>>>>>> >>>>>>> On Sat, Nov 1, 2008 at 3:01 AM, Chris Chabot <[EMAIL PROTECTED]> >>>>>>>> >>>>>>> wrote: >>>>>> >>>>>>> >>>>>>>> The good news here is that there's some progress being made on the >>>>>>>>> >>>>>>>>> REST+OAuth front, i got some great patches from Joseph Smarr for >>> >>> this >>>>>>>>> >>>>>>>> which >>>>>>>> >>>>>>>>> I'll be working to integrate into php-shindig these days. >>>>>>>>> OAuth was the last remaining big blocker, so that's really exciting >>>>>>>>> >>>>>>>> news! >>>>>> >>>>>>> >>>>>>>>> There's a few small remaining items in JIRA that needs to be fixed >>>>>>>>> >>>>>>>> before >>>>>> >>>>>>> php-shindig is completely ready, the tiered message bundle support >>>>>>>>> >>>>>>>> being >>>>>> >>>>>>> one, and checking the rss feed makeRequest processor format being >>>>>>>>> >>>>>>>> another >>>>>> >>>>>>> that I can think of from the top of my head, their smaller changes >>>>>>>>> >>>>>>>> though, >>>>>>>> >>>>>>>>> so I'm sure we'll be able to tackle and verify them inside of 2 >>> >>> weeks. >>>>>>>>> >>>>>>>>> There are a few improvements that would be nice to get in before >>> >>> then, >>>>>>>>> >>>>>>>> unit >>>>>>>> >>>>>>>>> tests for the new social api implementation being one of them, but >>>>>>>>> >>>>>>>> none >>>>>> >>>>>>> of >>>>>>>> >>>>>>>>> those things are worth considering release-blockers (it's been long >>>>>>>>> >>>>>>>> enough >>>>>>>> >>>>>>>>> already), however if someone has some time to contribute to that, >>>>>>>>> >>>>>>>> patches >>>>>> >>>>>>> are very welcome! :) >>>>>>>>> >>>>>>>>> How does Nov 17th sound for a possible release date? That would >>>>>>>>> give >>>>>>>>> >>>>>>>> us 2 >>>>>> >>>>>>> working weeks to finish stuff up >>>>>>>>> >>>>>>>>> Oh ps, does java-shindig support 3 legged oauth yet? I would hope >>> >>> we'd >>>>>>>>> >>>>>>>> get >>>>>>>> >>>>>>>>> that in for both the php and java versions in the 1.0 release since >>>>>>>>> >>>>>>>> that >>>>>> >>>>>>> seems essential for the full potential of rest / portablecontacts / >>>>>>>>> >>>>>>>> mobile >>>>>>>> >>>>>>>>> and server to server development. >>>>>>>>> >>>>>>>> >>>>>>>> >>>>>>>> It's semi-supported, but currently would require a lot of work from >>>>>>>> integrators. Louis and Brian were both looking into this recently, >>> >>> and >>>>>>>> >>>>>>>> I >>>>>>>> don't know how much work would be needed to finish it. >>>>>>>> >>>>>>>> >>>>>>>> >>>>>>>>> >>>>>>>>> -- Chris >>>>>>>>> >>>>>>>>> On Sat, Nov 1, 2008 at 12:57 AM, Ian Boston <[EMAIL PROTECTED]> wrote: >>>>>>>>> >>>>>>>>> I am ok for a branch, it would be good to check how the release is >>>>>>>>>> >>>>>>>>> going >>>>>>>> >>>>>>>>> to >>>>>>>>> >>>>>>>>>> be done with the mix of java and php, as I would assume this wont >>>>>>>>>> >>>>>>>>> just >>>>>> >>>>>>> be >>>>>>>> >>>>>>>>> a >>>>>>>>> >>>>>>>>>> tarball of the source tree ? >>>>>>>>>> >>>>>>>>>> Some blocker issues >>>>>>>>>> ------------------------ >>>>>>>>>> I have a minor concern and some advice needed on SHINDIG-562, >>>>>>>>>> which >>>>>>>>>> >>>>>>>>> is >>>>>> >>>>>>> the >>>>>>>>> >>>>>>>>>> blocker on the XML format for the RPC interface. >>>>>>>>>> >>>>>>>>>> There were a reasonable number of issues/bugs in the XSD that was >>>>>>>>>> >>>>>>>>> released >>>>>>>>> >>>>>>>>>> for 0.8.1 in the spec. >>>>>>>>>> These will be fixed in 0.9, but if I implement to the 0.8.1 it >>>>>>>>>> will >>>>>>>>>> >>>>>>>>> change >>>>>>>>> >>>>>>>>>> on the next release and wont be backwards compatible. >>>>>>>>>> >>>>>>>>>> I could take what I think the XSD for 0.9 would look like... but >>>>>>>>>> >>>>>>>>> thats >>>>>> >>>>>>> not >>>>>>>>> >>>>>>>>>> final, and this will need to replace the BeanXmlConverter and >>>>>>>>>> BeanAtomConverter beans that are currently wired in. There is a >>>>>>>>>> new >>>>>>>>>> >>>>>>>>> XStream >>>>>>>>> >>>>>>>>>> based converter that is lighter, faster and easier to use that >>> >>> would >>>>>>>>>> >>>>>>>>> replace >>>>>>>>> >>>>>>>>>> these 2 but is not currently wired into the module. >>>>>>>>>> >>>>>>>>>> Options: >>>>>>>>>> 1. Make SHINDIG-562 major and don't deliver compliant XML in the >>>>>>>>>> >>>>>>>>> release. >>>>>>>>> >>>>>>>>>> 2. Deliver to the 0.8.1 XSD as stands and expect to see major >>> >>> change >>>>>>>>>> >>>>>>>>>> 3. Deliver to a very early draft of the 0.9 XSD and expect some >>>>>>>>>> >>>>>>>>> minor >>>>>> >>>>>>> fixes. >>>>>>>>>> >>>>>>>>>> I would like to do 3, but that might be premature. >>>>>>>>>> --------------------- >>>>>>>>>> >>>>>>>>>> SHINDIG-561, SHINDIG-497 both blockers will be easy once 562 is >>>>>>>>>> >>>>>>>>> fixed. >>>>>> >>>>>>> --------------------- >>>>>>>>>> >>>>>>>>>> >>>>>>>>>> >>>>>>>>>> Ian >>>>>>>>>> >>>>>>>>>> >>>>>>>>>> On 1 Nov 2008, at 02:13, Kevin Brown wrote: >>>>>>>>>> >>>>>>>>>> Hi everybody, >>>>>>>>>> >>>>>>>>>>> >>>>>>>>>>> We've sort of waffled on release for a while now, and I'd like to >>>>>>>>>>> >>>>>>>>>> get >>>>>> >>>>>>> the >>>>>>>>> >>>>>>>>>> ball rolling again. Things seem pretty stable right now, but if we >>>>>>>>>>> >>>>>>>>>> don't >>>>>>>> >>>>>>>>> move to get a release the spec process is going to make releasing >>>>>>>>>>> >>>>>>>>>> soon >>>>>> >>>>>>> very >>>>>>>>>>> >>>>>>>>>>> difficult. >>>>>>>>>>> >>>>>>>>>>> So lets lock down and get released. >>>>>>>>>>> >>>>>>>>>>> We're pretty much fully 0.8.1 compliant (barring a few minor >>>>>>>>>>> >>>>>>>>>> issues). >>>>>> >>>>>>> We >>>>>>>> >>>>>>>>> have some of the 0.9 stuff already implemented, but we should not >>>>>>>>>>> >>>>>>>>>> consider >>>>>>>>> >>>>>>>>>> that "advertised" yet until it's really finalized and developers >>>>>>>>>>> >>>>>>>>>> start >>>>>> >>>>>>> expecting it. >>>>>>>>>>> >>>>>>>>>>> So, where do we stand? Can everyone agree on a branch and release >>>>>>>>>>> >>>>>>>>>> two >>>>>> >>>>>>> weeks >>>>>>>>>>> >>>>>>>>>>> from now? >>>>>>>>>>> >>>>>>>>>>> I'd like this to be the "1.0" release, with the next release >>>>>>>>>>> >>>>>>>>>> including >>>>>> >>>>>>> support for opensocial 0.9. I don't feel that it's a good idea to >>>>>>>>>>> >>>>>>>>>> align >>>>>>>> >>>>>>>>> our >>>>>>>>>>> >>>>>>>>>>> versions with the specifications versions, but we should include >>>>>>>>>>> >>>>>>>>>> something >>>>>>>>> >>>>>>>>>> in our version string or release notes that makes compatibility >>>>>>>>>>> >>>>>>>>>> obvious. >>>>>>>> >>>>>>>>> >>>>>>>>>>> Shindig will be 1 year old on December 12th (well, it's >>> >>> technically >>>>>>>>>>> >>>>>>>>>> a >>>>>> >>>>>>> few >>>>>>>>> >>>>>>>>>> weeks older than that but that's the oldest commit I can find in >>>>>>>>>>> >>>>>>>>>> SVN). >>>>>> >>>>>>> Surely we can get a release out before then! >>>>>>>>>>> >>>>>>>>>>> >>>>>>>>>> >>>>>>>>>> >>>>>>>>> >>>>>>>> >>>>>>> >>>>>> >>>> >>> > >
