RE: WebIDL v1 vs. v2

From: Sam Ruby [mailto:rubys@intertwingly.net] 

> I made a suggestion that hopefully should help webapps sort this out for themselves:
>
> http://lists.w3.org/Archives/Public/public-webapps/2014OctDec/0551.html

Right, it's possible that WebIDL v1 is so bad it's not even really possible to achieve REC with it (or with some reasonable subset of it, since e.g. reducing it to syntax leaves nothing to test). This may indeed sort itself out.

However, I'm concerned about the fact that this is even being considered, and think it would be a good area for the TAG to give guidance in. We've spent a lot of time working with spec authors to use newer concepts and IDL constructs (promises, iterables, maplike/setlike, binary data-related stuff, etc.) many of which are not only missing from WebIDL v1, but also actively incompatible with it (as Boris explains in the thread). It is IMO a dead-end spec and effort should be focused on WebIDL v2.

In other words, I'd like to head this off as quickly as possible to minimize confusion and eliminate any perception that WebIDL v1 should be used, referenced, or valued. Waiting for the process to reveal that through e.g. failure of implementations (which implement v2) to pass a v1 test suite during CR exit would be excessively painful and drawn-out.

I'd love for us to discuss this on an upcoming call or similar to see if the TAG at large generally agrees with my opinion, and if so we can present that opinion to webapps. Unfortunately the next one seems to be 4am my time but the call on the 11th might work...

Received on Wednesday, 3 December 2014 17:07:18 UTC