- From: Harry Halpin <hhalpin@w3.org>
- Date: Wed, 10 Dec 2014 17:09:05 -0000
- To: "Ryan Sleevi" <sleevi@google.com>
- Cc: "Harry Halpin" <hhalpin@w3.org>, public-webcrypto@w3.org
Note that the profile is non-normative. The point is to reduce, as you put it, vanity algorithms. > On Dec 10, 2014 6:26 AM, "Harry Halpin" <hhalpin@w3.org> wrote: >> >> During the CR transition call, the Director felt we were not explicitly >> clear enough about the status of the "browser profile", although he was >> happy to resolved the issue. I will thus have to link to one of the >> editorial notes in the spec's status with the proposed criteria below >> and a reference to the fact all "algorithms" are non-normative. The goal >> is to make sure implementers know right now that all algorithms are not >> supported cross browser, but should know after PR. >> >> In particular, the proposal was that we have two interoperable >> implementations for every algorithm listed in the spec body. However, >> the director is happy with algorithms being non-normative as long as >> there is a clear browser profile of interoperable algorithms. >> Interoperability will be determined by the test-suite. That is in line >> with WG discussions. However, if folks in the WG have an objection, >> please mention now. > > I still remain unconvinced we will be able to make meaningful progress or > normative requirement, given that both politics and local policy directly > affect what users experience, and so the availability of an algorithm > during a given browsing session with a given user agent will ALWAYS remain > non guaranteed. > > Our success criteria for continuing to include an algorithm in the spec > requires interoperable implementations, and I have no objection there. > > Discussing a browser profile as a way of trimming the list of > useless/vanity algorithms, I also agree there. > > But while I support exploring the possibility of discussing a browser > profile, and while I absolutely thing it is a TBD, I do not and cannot > guarantee that we will be supportive of a normative requirement for > something that is as much a deployment/user configuration issue as it is a > specification issue, nor when making an algorithm normative would inhibit > future security improvements. > >> >> We also estimated to the WG that we exit PR at estimated date of "March >> 12 2015". Again, usually 3 months is the minimum to get a unified >> test-suite together, and we're working in the fact that there is >> Christmas vacation soon. >> >> cheers, >> harry >> >
Received on Wednesday, 10 December 2014 17:09:11 UTC