On Jun 24, 2010, at 12:56 AM, Peter Cowburn wrote: > Hi Yannick > > On 23 June 2010 18:36, Yannick Torrès <yannick.tor...@gmail.com> wrote: >> 2010/6/23 Peter Cowburn <petercowb...@gmail.com> >>> >>> On 22 June 2010 19:08, Philip Olson <phi...@roshambo.org> wrote: >>>> Hello all, >>>> >>>> This proposal proposes the following: >>>> >>>> (1) Email the translation lists (doc-$lang) when a translation build is >>>> broken >>>> (A) Active translations daily >>>> (B) Inactive translations weekly >>>> (2) Email phpdoc@ when the English build is broken (daily) >>>> >>>> I don't expect many emails ;) Sound fine? Thoughts on improving this >>>> idea? >>> >>> Sounds great! My only "improvement" would be to perhaps run the build >>> test immediately after a commit (perhaps only email the author or a >>> selected list of folks?) if that is feasible. >> >> There are 2 points who let me know it's not a good idea : >> * The build process consume some memory and there is (can be) a "lot of >> commit" via the editor > > There is no requirement that the build process take place on the box > that runs the editor, is there? I've been playing around with farming > this sort of task out to "the cloud" but I'm not sure how folks would > feel with non-php.net hardware. Then again, what does it matter > between getting a broken build message soon after a commit, versus > getting one within 24 hours of it being broken... daily (or even less > frequent for less active translations) may well be "good enough".
It can be on any box, but personally I don't think complicating it is worth the resources and work involved but if it itches, please feel free to scratch it. I'm not sure what the best form of ping is, but people don't seem too keen on having the SVN server ping things with every commit, but maybe hooking something like identi.ca that everyone could use would actually save resources (and be more useful) in the end. Also, I hope people don't get the idea of not validating before svn commits.... ;) Regards, Philip
