In order to help the Tamarin team get their development of mozilla/js/tamarin moved off of cvs.mozilla.org, and to otherwise get going on Mozilla 2, we would like to have two repositories created today if possible: cvs-mirror what's now called trunk-mirror, the every six hour import from cvs-mirror.mozilla.org trunk the Mozilla 2+ trunk; we think it ought not be called "Mozilla 2" as it could outlive Moz2 (knock wood) Comments, objections? Any problems keeping this from happening? The import script is tracked in bug 376315. /be
Paul talked to me today, he was going to get together with me on Monday to get the cvs-mirror import going. I can work on this today, but Paul wanted to be around if we had any problems with the import. The trunk (moz2) repro - I can create that right away, but it won't contain any code? Who will push the initial code out to it? Should it be some sort of copy from cvs?
Yes, you can go ahead and create the moz2 repo. Somebody (probably me or graydon) will do the clone/push to populate moz2 from cvs-mirror.
You probably want to clone it server-side, to get the hardlinks.
Alright, I did a diff between a pull of cvs and a pull of the import clone, and got the following: [preed@preed-lx tools]$ diff -r /tmp/hg-current-remote-test/cvs-trunk-mirror/ /t mp/cvs-hg-imp-test/mozilla/ | grep -v '^Only in.*CVS$' | grep -v '^Only in.*cvsi gnore$' Only in /tmp/hg-current-remote-test/cvs-trunk-mirror/: .hg Only in /tmp/hg-current-remote-test/cvs-trunk-mirror/: .hgignore Only in /tmp/cvs-hg-imp-test/mozilla/browser: extensions Only in /tmp/cvs-hg-imp-test/mozilla/content/xml/content: public Only in /tmp/cvs-hg-imp-test/mozilla/docshell/resources: locale Only in /tmp/cvs-hg-imp-test/mozilla/js/tamarin/test/as3/Definitions: Regress Only in /tmp/cvs-hg-imp-test/mozilla/js/tamarin/test/as3/RuntimeErrors: SpecialT estCases Only in /tmp/cvs-hg-imp-test/mozilla/js/tamarin/test/as3/Statements/Exceptions: getStackTrace Only in /tmp/cvs-hg-imp-test/mozilla/rdf: resources Only in /tmp/cvs-hg-imp-test/mozilla/xpfe/communicator/resources/locale/en-US: mac Only in /tmp/cvs-hg-imp-test/mozilla/xpfe/communicator/resources/locale/en-US: unix Only in /tmp/cvs-hg-imp-test/mozilla/xpfe/communicator/resources/locale/en-US: win Looks like the directories in cvs-hg-imp-test are full of CVS and/or .cvsignore files/directories, so Hg refuses to import them (because to Hg, they're empty directories). Looks like we can create the mozilla2-trunk repo from this latest import.
Resolving this one, since both the mozilla2-trunk and the cvs mirror are now live.
Sorry for not reading this bug closer, but as Brendan said in his initial comment, we ought to not name the trunk something that can outlive mozilla 2, and now the name is mozilla2-trunk. Could we do a quick rename to simply "trunk"?
(In reply to comment #6) > Sorry for not reading this bug closer, but as Brendan said in his initial > comment, we ought to not name the trunk something that can outlive mozilla 2, > and now the name is mozilla2-trunk. > > Could we do a quick rename to simply "trunk"? Aravind and I ran this by Brendan and Benjamin on Friday; we picked mozilla2-trunk because to disambiguate from "the trunk," which canonically means the CVS trunk. The thinking is that it's easy to clone the repo when we're ready to move on, and name it something else when we're all working on the New Hotness.
Pardon me for reopening this again, but I still think we could do better. I mean, if we know the name isn't going to stick, why not take the time now to figure out what name could? If we don't like the name "trunk", how about "tip", or "head", or "master", "main"? None of which we've used much, or at all, for CVS, and they'd work just fine past Mozilla 2.
I defer to others with more experience with newfangled VCSes. My thought remains that "trunk" is well understood in our world, so if we think the metaphor works in hg, we should value continuity. Once we are happy with candidates among the people cc'd here, someone should post the recommended names to m.d.platform. /be
The more I think about it, the more I'm pretty sure that mozilla2-trunk is the correct name: 1) We have no plans for "past mozilla2" 1.1) If we did, I'd probably argue that we would clone a mozilla3-trunk repository as we've done this time 2) Renaming repos is relatively easy 3) We really need to avoid confusion between the 1.9 trunk and the moz2 trunk
I couldn't disagree much more than I do, but if I'm alone in that camp then so be it.
Get the proposal with the trade-offs about "trunk" and "mozilla2" name components noted fully posted to a (some?) newsgroups and let's see if someone has a better idea, or more people agree with jst. /be
Posted to m.d.platform and m.d.planning.
You posted to those places, but this is still a private (mozilla.org confidential) bug. Can it be opened?
What's the "standard" hg name for the main branch? I agree with jst that having a version number in a branch that conceptually lives forever isn't a good thing.
fwiw, I think I agree with jst, but I don't think the name matters too much. I would like to avoid "which trunk?" discussions, since I think it's unlikely people will bother to fully qualify the name in casual communication. I'll toss out "mozilla-central", per http://www.selenic.com/mercurial/wiki/index.cgi/UnderstandingMercurial#head-b451b95983ba488ef2b7b8802f93fd888460168e
I think this is degrading into a bike shed discussion: http://www.unixguide.net/freebsd/faq/16.19.shtml We can easily clone a repo (with full history) and call it something else when the new project comes along. It seems like we can easily rename a repo. The current name serves a purpose (disambiguation), and I see anyone making the claim that it's an unworkable name (like, say "cvs-trunk" or something like that). To be clear, if this is *that* big a deal, then let's re-initialize the repos, or rename them or do whatever we need to do. It'll take us into next week, still not having a repo set up *for actual use*. But to me, right now, the more interesting Mercurial-issue-of-the-hour to <strike>argue</strike> ponder over is how *often* we automerge from CVS, which a) is still being discussed and b) is theoretically blocked on this naming issue. I just don't think it's that big a deal, and it's precluding us from finishing setting up Hg so it's actually useful to anyone.
Another name to throw into the mix here would be hg-trunk, short, version independent, and makes it obvious what repo/branch it refers to.
> hg-trunk, short, version independent, obvious what repo/branch it refers to I'll second this, and I'll note it's somewhat analogous to cvs-mirror in its construction, and now I'll stop bugspamming.
Reassigning to preed temporarily, for him and jst to come up with a scripted driver for jst's mirroring tech.
Ok, aravind and I got jst's Hg access working finally. jst requested that the following be done: 1. clone cvs-trunk-mirror to mozilla-central 2. remove mozilla2-trunk 3. grant check-in permissions for jst to cvs-trunk-mirror and mozilla-central 4. nobody but jst, benjamin, preed should have write access to cvs-trunk-mirror, but it should be readable by all 5. mozilla-central should be read/write for all Hg users He would like this be done as early as possible (he requested tonight). Also, he wants to make sure that hg.m.o is being backed up.
clone cvs-trunk-mirror to mozilla-central - done remove mozilla2-trunk - done grant check-in permissions for jst to cvs-trunk-mirror and mozilla-central - done nobody but jst, benjamin, preed should have write access to cvs-trunk-mirror, but it should be readable by all - done mozilla-central should be read/write for all Hg users - done he wants to make sure that hg.m.o is being backed up - I will start backing up the repos tonight.
Can this be resolved now?
Yeah, we'll file followups as needed.