incubator-callback-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Brian>
Subject Re: [2/2] spec commit: Adding different bridge benchmarking to the Automated Mobile Spec Tests
Date Wed, 12 Sep 2012 12:15:04 GMT
personally don't care where the code lives either way the consistent
aggregate is what makes the cordova 'platform' valuable ---and havin a
bench in the same spot will help us better tune how we view changes
that could impact across each platform

On Tue, Sep 11, 2012 at 2:38 PM, Filip Maj <> wrote:
> I disagree.
> Bridge interface is identical across platforms (cordova.exec) and a
> platform-agnostic test working against the exec interface, comparing
> relative performance/correctness of each underlying implementation is a
> perfectly reasonable, and obviously useful, test to have around.
> On 9/11/12 2:37 PM, "Jesse MacFadyen" <> wrote:
>>These tests should live in their respective platforms. Exposing it as
>>an API gives away our sausage recipe, and no-one should ever care,
>>outside of the bridge developer.
>>  Jesse
>>On 2012-09-11, at 2:21 PM, Filip Maj <> wrote:
>>> Nice work on that Joe.
>>> I definitely support enumerating the bridge modes.
>>> I'm thinking this should be a standard field that platforms can override
>>> on a per-platform basis. In the top-level "cordova" module perhaps?
>>> On 9/11/12 6:26 AM, "Andrew Grieve" <> wrote:
>>>> On Tue, Sep 11, 2012 at 1:19 AM, Joe Bowser <> wrote:
>>>>> Hey
>>>>> Responses inline:
>>>>> On Mon, Sep 10, 2012 at 7:44 PM, Andrew Grieve <>
>>>>> wrote:
>>>>>> Hey Joe,
>>>>>> Wondering why make this into a jasmine test? Does it make the results
>>>>> more
>>>>>> easily captured?
>>>>> Yes, it also makes other known bugs glaringly obvious, like the
>>>>> numerous bugs with JS_OBJECT and the Callback Server. Now, instead of
>>>>> having to go through repro steps, I can just run this test.
>>>>> It also makes it easier to run a small amount of tests on a wide range
>>>>> of devices quickly instead of manually having to pick modes, and it in
>>>>> theory could work with the Continuous Integration that we're hoping to
>>>>> have in our office as well. I was doing testing on the HTC One X that
>>>>> arrived on my desk and my results looked different enough from the
>>>>> Galaxy Nexus that I wanted this. I was able to run through a
>>>>> half-dozen Android devices to see if the results on this end were
>>>>> similar to the ones that you had in the ticket.
>>>> That's awesome!!
>>>>>> Other thing I'm wondering is if this should use some JS reflection
>>>>> detect
>>>>>> the available bridge modes since they are different on iOS and
>>>>> non-existant
>>>>>> on others (mobile-spec tests are supposed to work on all platforms
>>>>> correct?)
>>>>> It probably would make sense for the bridges to be enumerated for
>>>>> readability.  So far, only iOS and Android have configurable bridges,
>>>>> right?  I think this test make sense here, but not added to the "Run
>>>>> All Tests" page.
>>>> Sounds good.

View raw message