Comment 3 for bug 593404

No worry Umang :)
No, we didn't decide that, here is the backlog of our conversation:

2010-08-05 12:53:05 umang didrocks, I didn't quite understand your comment in the merge proposal
2010-08-05 12:53:15 didrocks umang: ok so
2010-08-05 12:53:29 didrocks umang: I think with this scheme, we should forget about -public1,2,3…
2010-08-05 12:53:50 didrocks we should only use:
2010-08-05 12:54:09 didrocks <previous_rev>+r<rev>-<time>
2010-08-05 12:54:15 didrocks like
2010-08-05 12:55:01 didrocks 10.04+r54-1008051254
2010-08-05 12:55:10 didrocks for 10/08/05 at 12:54
2010-08-05 12:55:22 didrocks this way, someone can share multiple time the same commit
2010-08-05 12:55:23 umang didrocks, I've not got the time part as yet, but there's still "-public" in the code to remove "-public" if it is already there.
2010-08-05 12:55:26 didrocks (not sue if we want that)
2010-08-05 12:55:34 didrocks yeah
2010-08-05 12:55:49 didrocks in fact, I'm not sure that -public means something to people
2010-08-05 12:56:06 didrocks hence the idea of <time> which can be more meaningful
2010-08-05 12:56:11 didrocks I don't know, it's just a rough idea :)
2010-08-05 12:56:12 umang i.e. if it was already 10.04-public5, it should now remove -public5 and just become 10.04-r54-1008051254
2010-08-05 12:56:33 didrocks well, it should never adds public now, yeah
2010-08-05 12:56:41 didrocks what do you think?
2010-08-05 12:56:56 didrocks and no double -
2010-08-05 12:57:03 didrocks dpkg doesn't support it
2010-08-05 12:57:08 umang didrocks, I think i'm confused.
2010-08-05 12:57:17 umang didrocks, yes, no double -
2010-08-05 12:57:19 didrocks so 10.04+r54-1008051254
2010-08-05 12:57:21 umang :p my mistake
2010-08-05 12:57:27 didrocks on what are you confused?
2010-08-05 12:58:12 umang didrocks, I like what you are saying and agree about 10.04+r54-1008051254. I'm just confused about whether we should remove a -public if it is already in the version.
2010-08-05 12:59:15 didrocks umang: -public was just to enable people using quickly share multiple time
2010-08-05 12:59:16 umang didrocks, if you have my branch somewhere look at line 345 in internal/packaging.py
2010-08-05 12:59:25 didrocks I know your branch :)
2010-08-05 12:59:35 didrocks here, if they do:
2010-08-05 12:59:38 didrocks quickly share
2010-08-05 12:59:43 didrocks they get 10.04+r54-1008051254
2010-08-05 12:59:49 didrocks if they do quickly share again
2010-08-05 12:59:53 didrocks 10.04+r54-1008051256
2010-08-05 12:59:54 didrocks for instance
2010-08-05 12:59:59 didrocks so different numbers
2010-08-05 13:00:04 didrocks but that's probably a dummy idea
2010-08-05 13:00:07 umang didrocks, yes I like that.
2010-08-05 13:00:09 didrocks keep -public so
2010-08-05 13:00:22 didrocks 10.04+r54-public1
2010-08-05 13:00:24 didrocks then quickly share
2010-08-05 13:00:29 didrocks 10.04+r54-public2
2010-08-05 13:00:35 didrocks then commit + quickly share
2010-08-05 13:00:40 didrocks 10.04+r55-public1
2010-08-05 13:00:44 didrocks that makes sense
2010-08-05 13:00:57 didrocks your current branch does that, right? ^
2010-08-05 13:01:04 umang didrocks, no.
2010-08-05 13:01:35 umang didrocks, if someone was using an old version of quickly then did quickly share, they'd get 10.04-public5
2010-08-05 13:01:44 umang now when they upgrade to the new quickly
2010-08-05 13:01:55 umang and then quickly share
2010-08-05 13:02:16 umang they'd get 10.04-public5+r55
2010-08-05 13:02:34 didrocks hum
2010-08-05 13:02:34 umang line 345 removes -publicX completely
2010-08-05 13:02:38 didrocks no, so
2010-08-05 13:02:49 didrocks I would rather like we take a version from bzr
2010-08-05 13:02:51 didrocks so
2010-08-05 13:02:53 umang didrocks, so old and new users don't see -public at all
2010-08-05 13:03:18 didrocks well, you still need to be able to share twice without code change
2010-08-05 13:03:20 didrocks that can happens :)
2010-08-05 13:04:02 umang didrocks, but that's why we're going to put a <time> at the end, right?
2010-08-05 13:04:30 didrocks no, forget about my idea of time
2010-08-05 13:04:31 umang didrocks, I like the idea of <old>+r<rev>-<time>
2010-08-05 13:04:37 didrocks keep -public…
2010-08-05 13:04:43 umang didrocks, ok
2010-08-05 13:04:47 didrocks I think the time is more confusing finally
2010-08-05 13:04:57 didrocks speaking about it helped me to think more :)
2010-08-05 13:05:04 umang :)
2010-08-05 13:05:11 didrocks so, in a nutshell:
2010-08-05 13:05:15 didrocks we gave 10.04
2010-08-05 13:05:18 didrocks then, quickly share:
2010-08-05 13:05:24 didrocks 10.04+r54-public1
2010-08-05 13:05:28 didrocks quickly share again:
2010-08-05 13:05:32 didrocks 10.04+r54-public2
2010-08-05 13:05:37 didrocks then commit + quickly share:
2010-08-05 13:05:40 didrocks 10.04+r55-public1
2010-08-05 13:05:44 didrocks then, quickly release:
2010-08-05 13:05:48 didrocks 10.08
2010-08-05 13:05:50 didrocks oupss
2010-08-05 13:06:04 didrocks 10.08-0ubuntu1 (don't remember what we do :))
2010-08-05 13:06:07 didrocks or 10.08
2010-08-05 13:06:14 didrocks just keep what we have today
2010-08-05 13:06:19 umang didrocks, ok. I understand.
2010-08-05 13:06:29 didrocks great!
2010-08-05 13:06:43 umang :)
2010-08-05 13:07:36 didrocks thanks a lot!
2010-08-05 13:07:39 umang didrocks, sure.
2010-08-05 13:07:42 umang didrocks, just a question
2010-08-05 13:07:47 didrocks yeah?
2010-08-05 13:08:31 umang didrocks, with this format, we'll need to change the version number in upgrade.py not just in packaging (because you can't just do a version.split("-public") )
2010-08-05 13:08:55 umang didrocks, does that seem ok?
2010-08-05 13:09:23 didrocks yeah, making a change to transition people in upgrade.py sounds ok
2010-08-05 13:09:38 didrocks sync about the tree case, and test them:
2010-08-05 13:09:45 umang didrocks, sure I'll do that then.
2010-08-05 13:09:46 didrocks - people who have never shared already
2010-08-05 13:09:57 didrocks - people who have only packaged or released
2010-08-05 13:10:03 didrocks - people who has just shared
2010-08-05 13:10:26 didrocks and that launching the upgrade script on someone already transitionned shouldn't be harmful :)
2010-08-05 13:10:35 umang didrocks, sure.
2010-08-05 13:10:37 didrocks indempotent upgrade.py
2010-08-05 13:10:43 didrocks great :)
2010-08-05 13:11:11 umang didrocks, there's no hurry for this right? because I have no idea what the rest of this month is going to be like...
2010-08-05 13:11:21 didrocks umang: no hurry at all, take your time :)