On 16/01/14 02:14 AM, Vivien Nicolas wrote:
On 15/01/2014 18:59, Fabrice Desré wrote:
On 01/15/2014 09:54 AM, Vivien Nicolas wrote:
I also want to expose event loop lags if possible and make them a test
failure.
That looks very dependent on the performance of the test runner. How do
we take that into account?
Performance should vary depending on the test runner but in general the
event loop should not be blocked. So if the test runner is fast, there
will likely be no issue (while I'm pretty sure there will be some).
If the test runner is slow it will just exposed way more issue and the
tradeoff will depends on the UX we want to provide here.
Back when Project Snappy was just getting started I wrote a test harness
for desktop called Peptest for this purpose
(https://github.com/ahal/peptest). It never really ended up taking off
though mostly due to changing priorities.
It's always been a pipe dream of mine to get the harness working for
B2G, but it's so far down the priority list I don't think I'll be able
to do it anytime in the near or distant future. Currently some work
would be needed to get it working with B2G (e.g it currently uses
Mozmill's ui driver as marionette didn't exist back then), but switching
it to use Marionette shouldn't be *too* difficult.
_______________________________________________
dev-b2g mailing list
dev-b2g@lists.mozilla.org
https://lists.mozilla.org/listinfo/dev-b2g
