edit.js was killed in r91750
Aaron removed the the REL1_18 version of LQT for being "broken", then has brought in the version from 1.17wmf1, which is broken cause of a lack of dependancies
(In reply to comment #2)
Well the 1.18 version needed a schema change, has anyone looked into doing this
on the cluster?
I'll see what's needed now, and then look at scheduling the changes with Asher
I've also got a feeling that the version in 1.18wmf1 is now a version from 1.16wmf4
In r97086 I moved it from a 1.16wmf4 special to using REL1_18 directly..
Looking at the schema changes, there is nothing since September 2010, which is well before the 1.17 branchpoint
I'm marking this as a blocker for bug 29068 since there is at least one other wiki (ptwikibooks) where this bug will likely happen when the code is deployed.
(or there is a better tracking bug for this?)
I've forward-ported a bunch of changes I was making on another site, done a code review, and generally tried to get the code into a reasonable state. Some brief testing and it seems to work, I'd like it if folks could bang on it on the 1.18 cluster once the trunk state is deployed and let me know how it looks now.
My apologies for the mess of the current trunk state. Hopefully we'll have something nicer to deploy next time around :)
(In reply to comment #11)
Is there any actual bug in the version that's deployed? Reply works for me now.
If there isn't, then the severity should be downgraded.
Right now there is a symlink to handle the edit.js dependency. I'm not aware of anything still broken.
Sam and I planned out a deployment window for 20 hours from now (Wednesday, September 28, 20:00-21:00 UTC (1pm-2pm PDT)). Since Andrew has fixed up LQT on the trunk, this seemed like a sensible strategy.
If things aren't really broken, and assuming LQT in trunk isn't fully reviewed (I haven't checked that, so I'm guessing it isn't), then maybe we should call this off. That's not to say we couldn't deploy an update at some other point, but if things are at least nominally working, then it doesn't make sense to rush out a new version which could easily break other things.