Subject: Re: [boost] [outcome] Outcome v2 has reached stability, WG21 paper will be P0762R0
From: Thomas Heller (thom.heller_at_[hidden])
Date: 2017-07-31 15:25:16
On 07/31/2017 05:01 PM, Niall Douglas via Boost wrote:
> On 31/07/2017 15:04, Thomas Heller via Boost wrote:
>> On 07/31/2017 02:23 PM, Niall Douglas via Boost wrote:
>>> [...] and the
>>> "Boostification" cronjob script as the peer review wants a "pure Boost"
>> Where would one find the script and the result of the boostified
> As the announcement said, the boostification script has not been
D'oh ... missed that in the announcement... thanks for the clarification.
>> What's the rationale in having it run as a cronjob instead of
>> having the CI do the boostification upon a successful build?
> Outcome runs on both Travis and Appveyor. Each cannot know if the other
> succeeded. So a cronjob which already exists scans nightly the CDash for
> Outcome, and figures out the last commit with all tests passing on all
> platforms. It then updates master branch to that commit, and publishes a
> tarball. That same cronjob will eventually also Boostify Outcome into
> the Boost.Outcome repo if and only if all tests pass i.e. master branch
> got updated.
While the two CI systems don't know of each other, you can actually get
notified when a commit status update occurs. That would require some
service to listen on those status updates and logic to determine if all
builders passed (which is moot if you have testers not running on travis
or appveyor). So I guess a cron job is the only alternative for the lack
of proper CI support on this.