User:MarkAHershberger/Weekly reports/2011-10-24

It has been a few weeks since I wrote up a weekly report, so one is due to avoid this becoming a “monthly” report.


 * 1.18 Deployed
 * We successfully completed deployment of 1.18 to the cluster. However the number of regressions was a bit too large.  I watched WP:VPT to find a lot of these, so it really may not be that unusual.
 * Another benefit of watching WP:VPT and posting bug reports to was that I saw editors begin to post bug reports more frequently to Bz on their own.  I sense that the number of reports has stayed high this week but haven't had a chance to verify it.
 * In any case, I as I said after the triage of these bugs, I hope to start working with developers to create tests that will help us avoid these sorts of regressions in the future.
 * Another benefit of watching WP:VPT was gaining an idea of how to announce releases in the future. For future releases, we may try to use a site notice and ask editors to try a test wiki.
 * Someone had been pointing people who were asking “So what do we get from this” to the release notes. The average user found the “What's new?” page more readable.


 * NOLA Hackathon
 * I'm not sure how others feel about the hackathon, but I thought the NOLA hackathon was a success for me personally. I took advantage of the F2F time with core WMF devs to bang out some regressions and work with User:Reedy to sort out some shell bugs.  I'm not sure if Reedy got as much out of it as I did, but I understand better how User:RobLa thinks these things should be handled.
 * It was also fruitful in that I had a chance to meet with some developers supporting Microsoft SQL Server. This, User:Brion’s post to wikitech-l, and the above-mentioned tests will hopefully mean that the the next release of MW is that much more reliable.


 * 1.18 Tarball
 * Last week, User:Sumanah and I agreed that the next focus was the 1.18 tarball. This was reinforced by User:RobLa’s reminder at the end of the week.  What with the fair number of regressions from this time around, getting to tarball might take longer than anticipated.


 * FIXME progress
 * I've been hounding developer's with FIXME'd revisions to fix their code over the past couple of months. This has been super-gratifying in that we're down to less than a third of the FIXME's we had before.  Only two revisions are over a year old and 3/4 of the revisions are only a couple of months old.
 * Another good thing that has come from this is developer education. More develeopers have started using our code review tool… just as we're about to phase it out in favor of Gerrit.
 * The bad news is that Google has decided my Google-hosted email is sending spam, so some of my email's end up in people's Spam folders. This is a small price to pay, though.