<quote name="Rob Lanphier" date="2014-07-29" time="09:52:47 -0700"> > They are broadly responsible for the lifecycle of code from the point > that a developer is ready to check it in through its deployment on our > site, maintaining the processes and tools that reduce negative user > impact of site software changes while simultaneously making software > change deployment efficient and joyful.
Chris McMahon shared the below quote on the internal thread for this announcement, and I thought it was useful to share here as well: <quote name="Chris McMahon" date="2014-07-29" time="08:58:11 -0700"> > I think it's worth pointing out that RelEng is not only concerned with > releasing software early and often, but also concerned with releasing > software *safely*. You don't hear much about it, but stuff we also do: > > * Put in place and run all the linters, unit tests, qunit tests in Jenkins > * Deploy the master branch of all core and all extensions to beta labs > every three minutes > * Run automated browser tests in beta labs at least twice per day, and > analyze the results > * Do exploratory testing in beta labs > * Maintain the deploy tools like scap > * And manage the process within which all of these things are productive > > In Jenkins we find and fix code problems, for example with syntax and > structure. > > In beta labs we find and fix a number of sorts of problems: > > * configuration mistakes, like for caching or database. > * integration problems, for example when a change to VisualEditor makes it > stop working for MobileFrontend, or a change to Core breaks VE. > * regression problems, where a change in one part of the code unexpectedly > makes some other features stop working correctly. > > People sometimes ask me why the browser test builds are red so much. The > answer is that they are showing where changes and problems are. Red tests > give us information. > > So today we spend very little time in production "putting out fires", as > Andrew put it. Of course, we can't find and fix every problem, but I have > no doubt that our current practices and processes are saving Ops and Core > and Features engineers many frustrating hours every week. > > And speaking of practices and processes, having a Team Practices group in > place will be great. We have many interests in common. > > And if you're interested, I'm giving a short talk on the subject at > Wikimania: > https://wikimania2014.wikimedia.org/wiki/Submissions/Finding_and_fixing_software_bugs_for_the_Wikipedias -- | Greg Grossmeier GPG: B2FA 27B1 F7EB D327 6B8E | | identi.ca: @greg A18D 1138 8E47 FAC8 1C7D | _______________________________________________ Wikitech-l mailing list [email protected] https://lists.wikimedia.org/mailman/listinfo/wikitech-l
