**** BEGIN LOGGING AT Wed Nov 30 02:59:56 2005 Nov 30 10:54:43 koen, jbowler-zzz, do you have some time for me before I make a mess again :-) Nov 30 10:55:03 sure Nov 30 10:55:07 i'm trying to build from scratch from the mastermakefile Nov 30 10:55:27 and I get upon doing my initial checkout: Nov 30 10:55:28 monotone: misuse: branch org.openembedded.dev has multiple heads Nov 30 10:55:53 I'm a little bit reluctant to attempt a merge (and make a mess again) Nov 30 11:01:08 this is what I get: Nov 30 11:01:09 monotone: branch 'org.openembedded.dev' is currently unmerged: Nov 30 11:01:09 10189d45cc75641f13c4a9455d5eceaadb4a9816 jbowler@nslu2-linux.org 2005-11-29T20:06:43 Nov 30 11:01:09 affb16dd1c262c6f56c38e1ed230478f0aaa91d7 jbowler@nslu2-linux.org 2005-11-29T18:03:16 Nov 30 11:06:22 koen, would it be ok to do a mt merge -b org.openembedded.dev ? Nov 30 11:07:06 eFfeM: have you spoken to rwhitby yet? Nov 30 11:07:20 I think jbowler said something about not trying to merge Nov 30 11:07:23 jelle, no Nov 30 11:07:48 ok, he said he had some trouble fixing things after you propagated to openembedded: Nov 30 11:08:01 (10:35:09) rwhitby: if someone sees eFfeM, let him know that I've fixed his mess in org.openembedded.dev - it seems he someone propagated from org.nslu2-linux.dev to org.openembedded.dev (who knows how, cause the Makefile simply cannot do that). Nov 30 11:08:01 (10:35:24) rwhitby: He should send a very apologetic email to the OE list.   Nov 30 11:08:11 (that was yesterday I think) Nov 30 11:08:32 so be a little careful, don't want to mess up things twice. At least not by accident. Nov 30 11:08:36 ok, then I won't, actually I had the impression from jbowlers msg to the group that things were fix Nov 30 11:08:56 jelle, that was why I was asking before doing anything Nov 30 11:09:11 sounds good. I dont know any details, just bringing the bad news ;) Nov 30 11:09:54 I did get some email from jbowler-zzz and actually my tree was wrong from the very beginning and he istructed me to start with a fresh tree (which was what I was trying to do) Nov 30 11:10:37 and yes, I sincerely apologize for the mess I made. Wish I knew what caused it because then I'm sure not to make it again Nov 30 11:11:06 would be good. But I think you should address the OE list, since that's where it went wrong. Nov 30 11:11:09 it was definitely not my intention to mess up Nov 30 11:11:14 of course not. Nov 30 11:11:33 I thing rwhitby was mostly surprised that it could be messed up. Nov 30 11:12:00 how do I address the OE list? I don't know that one? I only know the nslu2 wiki, the yahoo groups on nslu2 and the irc channels Nov 30 11:13:02 actually if someone is knowledgeable on how to analyse this: I still have a copy of that faulty db on my system (I've put it aside as an insurance against loosing work that for whatever reason didn't make it) Nov 30 11:17:51 I think there might be a yahoo group openembedded or something. Otherwise try to figure it out with rwhitby-away, he'll probably wants to know what happened too, to avoid it. Nov 30 11:18:47 I'm just peeking in the log of my old tree and I think I know the cause, will report my findings here in some minutes Nov 30 11:36:22 koen, jbowler-zzz, rwhitby-away and all others that might have suffered from my actions or that are just interested to know what I did wrong: Nov 30 11:36:37 I've been researching why I made such a mess. Nov 30 11:36:37 The root cause is that when I got write access I was unaware that Nov 30 11:36:37 there are actually two databases. Nov 30 11:36:37 Probably I messed up by giving mt commands in the wrong directory. Nov 30 11:36:37 In the log I see that some files are added when I was in the wrong dir: Nov 30 11:36:47 Branch: org.openembedded.dev Nov 30 11:36:48 Added files: Nov 30 11:36:48 openslug/openembedded/packages/cdstatus/cdstatus-0.96.05.bb Nov 30 11:36:48 openslug/openembedded/packages/cdstatus/cdstatus-0.96.05/cdstatus.patch Nov 30 11:36:56 Also (one way or another) I managed to get a wrong MT/options: Nov 30 11:36:57 [slug@Woonkamer MT]$ cat options Nov 30 11:36:57 branch "org.openembedded.dev" Nov 30 11:36:57 database "/home/slug/OpenSlugHead/monotone/nslu2-linux.db" Nov 30 11:36:58 key "" Nov 30 11:37:12 No idea how I managed to do that, I definitely did not change that file on Nov 30 11:37:12 purpose. Also no idea why this went on undetected for such a long time. Nov 30 11:37:12 My sincere apologies for this. Nov 30 11:37:24 Actually perhaps the lesson for this is that we need an introductory document Nov 30 11:37:24 for newbe's like me. Don't know whether it is possible to put restrictions on what one can do where. That could also help preventing this from happening again. Nov 30 11:37:24 Anyway, I hope no-one is too mad at me. I think I've learned my lesson. Nov 30 11:37:24 I have to go now, back in 1.5 hr or so. Feel free to leave a msg Nov 30 12:15:00 eFfeM-away: I'll make a note of your lines, and drop it to rwhitby-away when I see him, just to be sure. Nov 30 13:29:36 eFfeM-away: as jelle said, I was mostly surprised at how the weird propagation of org.nslu2-linux.dev to org.openembedded.org happened. I understand that it was truly an accidental occurrence. Nov 30 13:30:48 i think monotone messed with the ancestor IDs Nov 30 13:31:27 eFfeM-away: best way to make up for the accident is to write it up on a wiki page so that we can avoid it in the future. Nov 30 13:33:39 koen|bar: yeah, it is very weird. I cannot see a way that the nslu2-linux stuff could be added to the openembedded tree by accident. The other way around is not too hard to do (if you're in the top-level and mt add something, then you can end up with openembedded stuff being accidentally added to the nslu2-linux branch), but I just can't see how it could have happened with normal monotone add commands without there either being a bug in mon Nov 30 13:43:49 koen|bar: from the OE point of view, has it all been repaired ok? Nov 30 13:44:08 (i.e. is there anything else we need to do to recover from the error?) Nov 30 13:47:18 everything looks ok Nov 30 13:55:33 rwhitby-away, koen|bar, I'll try to scribble a wiki page with some info that I learned through koen or the hard way Nov 30 13:56:16 won't do that today though, perhaps tomorrow Nov 30 13:56:51 i'll probably ask you guys to review it, to see if I wrote nothing silly Nov 30 13:57:45 and if you guys feel the oe community deserves an apology from me, just let me know where to send it to. Nov 30 13:58:04 (actually I don't know too much about the underlying things like oe, cia etc etc) Nov 30 13:58:41 oops, unexpected interrupt, but I'll check for other msgs before I go to zzzz Nov 30 13:59:32 eFfeM-away: check with koen|bar about the apology. We may have fixed it in time such that not many people were affected. Nov 30 14:34:23 eFfeM-away: no one using the OE databases saw these revisions because the OE servers do not export the required ancestors (which are in the org.nslu2-linux.dev branch) Nov 30 14:38:57 What seems to have happened (and this is a 'best guess') is that your root MT/options branch got changed from org.nslu2-linux.dev to org.openembedded.dev but your MT/revision still referred to revision e14fbd1c4191cdab47cc012f4ddbad43e71645b2 in the org.nslu2-linxux.dev branch. Nov 30 14:40:48 Then on 2005Oct20 you commited ef52fb8de8658c30581cea89db5730a944f8a8bc, which added 'openembedded/packages/pvrusb2-mci/...' stuff with an ancestor of that revision, but no branch Nov 30 14:44:01 Then on 2005Nov05 70f8ac179abbb777f1bec20edf02530820b33e92, with branch org.openembedded.dev, then this rev was merged with the current head of the org.openembedded.dev branch. Nov 30 14:46:11 At this point the result was that org.nslu2-linux.dev and org.openembedded.dev got added together - the result contained all the files in either plus the 'openembedded/packages' stuff from the original problem revision. Nov 30 14:48:06 So far as I can see the original error was the change to the top level MT/options, and everything resulted from this. Nov 30 14:50:19 rwhitby-away: the OE guys can't see this problem because their databases don't have org.nslu2-linux.dev (see the OE.db snapshot) and, unless they pull that they never get the problem revisions (because they have unsatisfied dependencies.) Nov 30 14:52:20 So their head is always the clean line which excludes the problem revisions and any of the merges we did afterward. Nov 30 14:54:31 I've rebuilt all my databases and changed my "pump" script to work with an individual database for each branch - so it can no longer distribute any revision with this problem, I still haven't worked out how to fix it on nslu2-linux.org though. Nov 30 14:58:04 Hum, no, the stray revision has managed to creep into my new system too... Nov 30 15:21:28 koen: what head revision do you have on org.openembedded.dev? Nov 30 15:40:59 ebcf7913b985f1954e5315bca0dfc1c50bb2c940 Nov 30 15:47:03 Yep, NAiL tried to merge the problem revision in again - doesn't work, it's only visible to someone who has explicitly synced to monotone.nslu2-linux.org Nov 30 16:01:38 jbowler-zzz, thanks for explaining this to me. If there is anything I can do to help cleaning this up let me know Nov 30 16:01:45 (but for now I'm calling it a day). Nov 30 16:02:19 i'll try to come up with a wiki page for beginning developers ... Nov 30 16:03:00 cu all **** ENDING LOGGING AT Thu Dec 01 02:59:57 2005