While going through the bugs that were marked as 1.20 tarball blockers today, most seemed to have been left to rot -- that is, they didn't really deserve the tag since people were not motivated to fix them and the people who marked them 1.20 tarball blockers didn't put in the effort to find someone to fix them.
Sometimes, that person was me, so this shouldn't be seen as overly harsh criticism.
I did see a couple of exceptions:
* https://bugzilla.wikimedia.org/38273 -- Tidy occasionally isn't executed
Seems to be happening somewhat regularly since July on different WMF sites. Some effort to track it down has been made, but no solution has been found yet. I've a feeling this bug would bite third-party users of MW if we released a tarball with it.
* https://bugzilla.wikimedia.org/35894 -- Reports of secret key generation "hanging" on windows`
This is a bug that will really affect only new tarball users on Windows. While this probably isn't a lot of people, I would still like to see it fixed before a tarball is released. MaxSem had some ideas and I'll talk to him about how feasibly he can implement them.
Otherwise, there are several patches mostly by Krinkle that need to be merged for a 1.20 tarball:
https://bugzilla.wikimedia.org/31676 - ResourceLoader should work around IE stylesheet limit https://bugzilla.wikimedia.org/38158 - jquery.byteLimit sometimes causes an unexpected 0 maxLength being enforced https://bugzilla.wikimedia.org/40448 - Replace mediawiki.legacy.mwsuggest with SimpleSearch. https://bugzilla.wikimedia.org/40500 - ResourceLoader should not ignore media-type for urls in debug mode https://bugzilla.wikimedia.org/35923 - Followup to bug 11374 - tweaks to mediawiki.action.history.diff.css https://bugzilla.wikimedia.org/40498 - ResourceLoader should not output an empty "@media print { }" block.
I still plan on being in #wikimedia-dev at UTC1100 on Tuesday, October 2 (See [0] for timezone conversion) for the triage, but I'll probably be spending that time putting in merge requests for the above issues (unless someone beats me to it).
Also Tuesday, I'll make a new tarball with the code merged as my final release candidate for 1.20. Sam has said that because we're doing a review-first model with Gerrit people should feel more liberty to put merge requests in, so have at it! [1]
Mark.
[0-short] http://hexm.de/lr [0-long] http://www.timeanddate.com/worldclock/meetingdetails.html?year=2012&month=10&day=2&hour=11&min=0&sec=0&p1=240&p2=44&p3=37&p4=136&p5=179&p6=224
[1-short] http://hexm.de/lw [1-long] https://gerrit.wikimedia.org/r/#/q/status:open+project:mediawiki/core+branch:REL1_20,n,z