[00:31:00] --- deason/gmail has left
[00:44:05] --- Russ has left: Disconnected
[02:48:04] --- lars.malinowsky has become available
[04:43:03] --- lars.malinowsky has left
[07:21:50] <Simon Wilkinson> Just spent an hour staring at RX packet traces, wondering why so many packets had been dropped. Then realised I'd managed to commit with the "intentionallyDropPackets" debugging feature turned on. d'Oh.
[08:27:25] --- jaltman/FrogsLeap has become available
[08:33:20] --- jaltman/FrogsLeap has left: Replaced by new connection
[08:33:21] --- jaltman/FrogsLeap has become available
[08:51:56] --- jaltman/FrogsLeap has left: Disconnected
[09:06:05] --- jaltman/FrogsLeap has become available
[10:05:47] --- Russ has become available
[13:51:10] --- deason/gmail has become available
[14:11:33] <jblaine> k, I'm confused again
[14:12:10] <jblaine> made gerrit 4850 from local branch man-html-css
[14:12:34] <jblaine> left that branch, came back to it just now, made another small change related to 4850
[14:12:43] <jblaine> git commit --amend -a
[14:12:57] <jblaine> I try to: git push ssh://gerrit.openafs.org/openafs.git 46da50f73729ff712bd66b753d21e3427d743a95:refs/changes/4850
[14:13:17] <jblaine> 46da50f73729ff712bd66b753d21e3427d743a95 being the commit that went into original 4850
[14:13:26] <jblaine>  ! [remote rejected] 46da50f73729ff712bd66b753d21e3427d743a95 -> refs/changes/4850 (no changes made)

[14:13:29] <kaduk@mit.edu/barnowl> refs/changes/50/4850
[14:14:07] <kaduk@mit.edu/barnowl> Also, if you want to push the updated commit, you should use the SHA1
of the new commit (or just 'HEAD'), not the old one.
[14:14:51] <kaduk@mit.edu/barnowl> I do not know of a robust way to go from the gerrit ID 4850 to that
path refs/changes/50/4850 other than visiting gerrit over HTTP and
looking at the download/etc. URLs it produces.
[14:15:14] <jblaine> git push ssh://gerrit.openafs.org/openafs.git HEAD:refs/changes/4850 worked
[14:15:32] <kaduk@mit.edu/barnowl> It is probably much easier to install the change-ID commit hook and
just push to refs/for/master, though.
[14:15:42] <jblaine> this is based on: http://wiki.openafs.org/AFSLore/GitDevelopers/ section "Revising your change"
[14:15:49] <kaduk@mit.edu/barnowl> Oh, huh.  Maybe I am thinking about fetching or something, or maybe it
was fixed in a gerrit update.
[14:16:11] <jblaine> I'm using the change hook
[14:16:12] <kaduk@mit.edu/barnowl> But anyway, it is expected that the commit will be rejected if it is
identical to the previous one.
[14:16:26] <jblaine> So if I'm using that, I don't need to worry about telling gerrit the specifics, eh?
[14:16:42] <kaduk@mit.edu/barnowl> That's the theory.  It works pretty well for me.
[14:16:54] <jblaine> thanks
[14:17:19] <kaduk@mit.edu/barnowl> (It only seems to cause trouble when trying to cherry-pick a commit
from master onto a stable branch, and one forgets to generate a new
change-ID.)
[14:57:16] --- jblaine has left: Lost connection
[14:57:16] --- meffie has left: Lost connection
[16:09:39] <Simon Wilkinson> In general, you should never need to use the explicit refs/changes/* paths. The GitDevelopers guide should probably be updated to remove the references to them.
[16:09:52] <deason/gmail> those instructions are from before we used change-ids all the time, I think
[16:10:04] <deason/gmail> now I don't even think it's possible to submit something without a change-id in the commit message
[16:10:16] <Simon Wilkinson> Yeah. I think they actually predate the version of gerrit that supported ChangeIds.
[16:10:40] <deason/gmail> in most circumstances, you just want to say HEAD:refs/for/master (or refs/for/openafs-stable-1_6_x or whatever)
[16:10:45] <Simon Wilkinson> Yeah, I turned that on a while back - we require ChangeIds with all commits. Sadly gerrit's handling of ChangeIds across multiple branches isn't as robust as it could be.
[16:12:02] <deason/gmail> by the way, if we upgrade to 2.2.x allegedly the in-gerrit diff viewer will work in opera again, which would be really neat
[16:52:59] <summatusmentis> wait, really? refs/changes is old? totally didn't know that
[17:25:22] --- mfelliott has left
[20:48:55] --- andersk has become available
[23:48:12] --- deason/gmail has left