Front page | perl.perl5.porters |
Postings from July 2010
On Tue, 13 Jul 2010, Tim Bunce wrote: > On Mon, Jul 12, 2010 at 09:22:27AM +0100, Nicholas Clark wrote: >> On Sun, Jul 11, 2010 at 10:06:07PM -0400, David Golden wrote: >>> My question is sparked by Karl's recent work on unicode, but it's not >>> specific to him. >>> >>> Now that blead is getting released on a monthly cycle, I'm not clear >>> on whether there is consensus that sweeping experimental changes have >>> a place in blead. > >> As-is, I don't think that feature branches would work, given that >> >> a: No smoker is, or knows how to, smoke arbitrary feature branches, so they're >> never going to get smoke tested. Few to no individuals touch them either. >> So how are we going to know that they pass muster? > > Rather than try to teach smokers how to smoke *arbitrary feature branches* > how about reserving a single git label, say "alt-smoke-pumpkin", that > smokers could optionally also do smoke testing on. > > That label could then be "passed around" and applied to any branch that > would benefit from smoke testing. That would be fine with me. I could set up a Test::Smoke instance running with git masquerading as rsync. >> b: For whatever reason(s), we don't have enough people reviewing and applying >> patches currently. On the assumption that more branches means more work >> to maintain, surely this is going to spread the existing volunteers even >> thinner. > > Doing the above would mean anyone working on a feature branch could get > the benefit of smoke testing *before* asking p5p for a detailed review. If people "steal" the tag frequently then smoking could be skipped for some people. My Win32 smoker has a 4 hour cycle but my Ubuntu smoker runs more around 18 hours. -- George GreerThread Previous | Thread Next