Re: "Pre-Final" Draft of Coremob - 2012 and dial-in details for telecon to advance to "Final"

Thanks for your input, Giri.

See my comments inline.

On 1/29/13 5:54 PM, "Mandyam, Giridhar" <mandyam@quicinc.com> wrote:

>Thanks to the editors for all of their work in coming up with this latest
>version.  My apologies if I am repeating comments, or raising issues that
>have already been addressed.
>-Giri Mandyam, Qualcomm Innovation Center
>
>General comments on the latest draft:
>
>1) Section 1, Introduction.  "Compile related conformance test suites."
>I realize this is simply quoting from the charter, but I think the
>charter should be slightly modified to read "Compile related conformance
>and performance test suites."

We should look at modifying the charter when we recharter the CG.

>2) Section 1.5, Aspirational nature of the document.  "quality of
>implementation" could be better defined, particularly what
>"implementation" means.  I think "implementation" should consider the
>browser + OS + CPU.

Point taken. Could you provide better wording?

>3) Section 2.4, Find the nearest subway station.  I think we should put
>in a requirement on geolocation accuracy.  Currently the Geolocation API
>spec does not provide guidance on how to implement the enableHighAccuracy
>flag.  I think that implementations that ignore this flag or do something
>unintended (e.g. use Cell ID when this flag is set and a GPS measurement
>is available) should not be considered conformant.  This however means
>adding conformance requirements to the Geolocation spec that do not exist
>as far as I can tell.

Adding normative requirements to spec is out of scope of this group and
should be handled in the relevant WG.

>4) Section 2.5, Read an online magazine.  "While on the subway, she is
>able to read the daily news, watch a video, ...".  What if the video is
>DRM protected and requires network access?  Also, I think Req. 10 should
>be applicable in this case.

The CG as a whole agreed not to tackle DRM for now.

>5) Section 2.6, View a regular Web site.  Is there an implicit
>requirement that the browser should be able to change the header User
>Agent string (i.e. to avoid retrieving the mobile version of the bank web
>site)?

No. Nothing in the use case says the user agent sniffing is used to serve
mobile specific content. The mobile website could well be hosted on a
different subdomain.

>6) Section 3.4, Multimedia.  Req. 18.  Regarding the 30 fps "must"
>requirement, this is dependent on how we define "implementation" (see my
>2nd comment above) and under what conditions the fps measurement is
>taken.  Moreover, how fps is measured would need to be defined.  I think
>in practice the best you can do with this requirement is make it a
>"should".

That also seems like a fair point. How about rephrasing the requirement
like so:

"It MUST be possible to smoothly draw multiple animated sprites in full
screen mode (e.g. at 30fps)."

Thanks,

--tobie

Received on Wednesday, 30 January 2013 06:47:17 UTC