- From: Harald Alvestrand <harald@alvestrand.no>
- Date: Tue, 07 May 2013 20:28:10 +0200
- To: public-media-capture@w3.org
On 05/07/2013 08:01 PM, Martin Thomson wrote: > On 7 May 2013 09:54, Cullen Jennings (fluffy) <fluffy@cisco.com> wrote: >> As far as I can tell, sourceID are more or less opaque identifiers. So can we move them to be integers to help make it easier to process constraints. It would be fine if this integer was created by doing a hash of some string. >> >> Note … I don't care much one way or the other if we do this but seems like not all bad idea …. > How big an integer? I'm thinking that the JavaScript numeric types > don't have enough bits for what I was planning to do. > > How does this make constraint processing easier? By removing string > (i.e., enumerated) types from the set of permissible constrain types? > I am wondering if it's equally effective (or ineffective) in terms of "avoiding the unconstrained string" to specify that these constraints must be valid identifiers out of some identifier space (in this case, the space of valid device identifiers). API code has to process enums (which are strings in Javascript) against a list of valid values, so in principle the same thing should be reasonable for device identifiers. Either is a pushback against "all new constraints will be unconstrained strings". Neither is terribly strong.
Received on Tuesday, 7 May 2013 18:28:39 UTC