On Sat, Oct 25, 2014 at 05:48:01PM +0200, Lodewijk andré de la porte wrote:
Standards should merely enable interoperability, thus create choice, and through that choice must come features, and the want for features will enforce those standards.
Related: who's up for swapping layer 3 altogether?
All I ask is that whatever comes out of it have an unambiguous BNF definition for its message format, without <prose-val> rules if at all possible. (See also: "enable interoperability".) Bonus points for defining the protocol itself as a finite state machine. WebCrypto is a shitshow in large part because the people at its wheel perceive a need for JavaScript programmers to make decisions about what cipher mode to use. They're dead-set on forcing developrs who write Javascript because C is hard to make low-level decisions that affect the reliability of their code in profound and non-obvious ways, and refuse to understand that this approach never ends well. Cheers, --mlp