On 2012-09-29 10:54 PM, Justin Dolske wrote:
On 9/29/12 1:14 PM, Justin Lebar wrote:
As part of this debate, I calculated that for patches with a
reasonable chance of success, pushing to m-i and failing actually
saves resources.
Hmm. Veering off on a bit of a tangent...
Perhaps we could find some quick load reductions by coalescing pushes
that have had successful try runs. EG, If 20 changesets have
successfully had try runs, doing 20 more (full!) runs on mozilla-inbound
is wasteful.
For example: We could create a mozilla-inbound-tried repo, which only
triggers builds once every N changesets, at which point it gets merged
(assuming greenage). Perhaps limiting it to changes will well-defined
impact (ie, where breakage makes the culprit obvious) would also be
useful, especially where a change passes try but introduces a new orange.
Creating more mozilla-inbounds creates more problems than it solves.
I've been preaching that for quite some time, perhaps it's time for me
to consolidate my argument into a blog post or something...
Ehsan
_______________________________________________
dev-platform mailing list
dev-platform@lists.mozilla.org
https://lists.mozilla.org/listinfo/dev-platform
