On 4/17/14 12:14 PM, "OmPrakash Muppirala" <bigosma...@gmail.com> wrote:
>On Thu, Apr 17, 2014 at 12:09 PM, Alex Harui <aha...@adobe.com> wrote: > >> OK, so what do others think? I'm totally fine with taking my Azure VM >>and >> turning it into the 'official' Flex CI server. We can try a 'medium' >> instance and see if it falls behind as we add more checkintests on the >> FlexJS side, and if we need to upgrade to 'large' I'll pay for it >>myself. >> >> It just seems like it will take less of our time than working with >>Apache >> Builds. For FlexJS we are probably going to want to install and >>maintain >> a variety of browsers. Right now we've disabled the FlexJS checkintest >>on >> Apache Builds because it didn't run and there was no simple way for me >>to >> debug it. I assume we really do want to run some sanity tests for each >> build, yes? >> >> >If this is within ASF's rules, I am all for it. Makes our lives so much >easier. IMO, it is within the rules. > >Do we want to at least use the ASF build server for providing official >nightly dev builds? If we take care of running checkintests on the VMs, >then we can do the builds on Linux as well, right? Then we wont need to >restrict our jobs on Windows slaves which seem to have the most problems >on >ASF build servers. IMO, we don't need that unless you think folks care about the URL they pull the nightly from. The nightly is not an official release and plenty of caveats already apply. And if the nightly build cannot be failed by a failing checkintest, then we'll end up with folks getting some bad ones. -Alex
