W3C home > Mailing lists > Public > public-device-apis@w3.org > December 2009

Re: First stab at the Capture API (aka Camera API)

From: Dominique Hazael-Massieux <dom@w3.org>
Date: Wed, 02 Dec 2009 11:41:43 +0100
To: "Tran, Dzung D" <dzung.d.tran@intel.com>
Cc: public-device-apis <public-device-apis@w3.org>
Message-ID: <1259750503.30579.623.camel@altocumulustier>
Le mardi 01 décembre 2009 à 19:50 -0800, Tran, Dzung D a écrit :
> It is checked in to CVS as stated below. There is some formatting
> problems, which I am strap for time to figure out right now. 
> http://dev.w3.org/2009/dap/camera/Overview.html

Some  thoughts and questions on the current proposal:
• if I understand it correctly, the current API assumes that the capture
activity is done separately from the Web environment — in particular,
the fact that the callbacks return with an array of results, and the
existence of the limit parameter, suggests that all the expected
materials is returned at once; this seems to entirely prevent a browser
from embedding the capture activity inside the viewport (where
presumably the capture could be done in several rows);

• it’s not clear if the limit parameter is binding — can a developer
assume that the MediaArray she gets from the callback function will have
a length inferior or equal to limit?

• the limit and duration parameters are non-optional; it’s not clear how
you would allow the user to capture as many and as long media as he
would need;

• duration is currently a double, counted in seconds; I think it would
make more sense to count in miliseconds (or microseconds?) and possibly
make it an unsigned long long

• I think MediaData should have an (optional?) timestamp parameter that
tells the time and date at which the capture was made — this enables
interesting use cases (e.g. sorting a sequence of media by the time they
were taken; geo-tagging media by relating the timestamp to the ones
recorded by the geolocation API)

• I’m not sure there is much points in defining MediaArray; I think we
ought to refer directly to sequence<MediaData>; in any case, it should
be renamed to MediaSequence if we keep it (array and sequences are
distinct in WebIDL)

• the current API hangs off of navigator.device (at least in the
example) — I have added an an issue in the document that this interface
would need to be defined somewhere

Dom
Received on Wednesday, 2 December 2009 10:41:54 GMT

This archive was generated by hypermail 2.2.0+W3C-0.50 : Wednesday, 9 May 2012 00:14:02 GMT