On Fri, Nov 23, 2007 at 02:40:06PM +0100 I heard the voice of Richard Levitte, and lo! it spake thus: > > One way, a bit complicated but still workable: I would simply have > kept hacking and committing into free.lp.se:X.ctwm in a separate > database (ctwm.fullermd) [...]
Mmm. The downside of that is making a repo for every new [concurrent] branch. That's awful heavy-weight. > Even simple, though, is to just keep one database, but then hack > away as needed, maybe in a separate workspace, commit whenever you > feel like, and when you're done, pull from guardian, merge and push. But the downside of THAT is that I can only do one thing at a time, have to run it to completion before touching anything else, and a poorly-timed push will end up putting that incomplete stuff out in the world as another head on the branch to mess with other people. I don't much like that either :| How do other mtn-using projects do this? Presumably, cert-less revisions are a sufficiently Bad Thing(tm) that it's not an intended result (otherwise, -viz would treat them a little better). Certainly working one thing to completion at a time defeats the whole point of branching. I don't believe the standard workflow involves creating a new repo for every transient branch. And I have a really hard time believing every mtn-using project just grows an ever-increasing number of branches. That would be nuts; I've got plenty of rather small projects where that list would run to 50 or 80 branches, with all but 1 or _maybe_ 2 of them being defunct and pointless. And some of them would be _completely_ worthless, because they were branches that were made to try something that turned out to be a bad idea, and so were discarded without ever being merged. And that leaves completely aside the nastiness of globally-unique naming of the branches, avoiding which is one of the big plusses of the D in DVCS. If it's gonna be globally visible, or even eternally locally visible, I could never make a X.ctwm.compiler_errors; it would have to be X.ctwm.fullermd.compiler_errors.200710 or something even more painfully unwieldy and potentially misleading like that. -- Matthew Fuller (MF4839) | [EMAIL PROTECTED] Systems/Network Administrator | http://www.over-yonder.net/~fullermd/ On the Internet, nobody can hear you scream.