--- Begin Message ---
On Fri Sep 11 10:55:10 2009, Alexey Melnikov wrote:
Any thoughts on this? I am concerned that this changes the
capability processing logic in the client from a simple token
lookup.
Polymer has always assumed that capabilities are either:
FOO
Or:
FOO=BAR FOO=BAZ
And not a mixture.
In fact, they're mapped internally to a dictionary construct, with
the FOO as key, and a list of the "sub capabilities" as the value:
{ "FOO": [] }
{ "FOO": ["BAR", "BAZ"] }
So if there's a UTF8 and a UTF8=ALL, that's likely to break things
for me, at least.
What I don't know is whether other programmers chose to treat
caoability strings this way or not - it's certainly mandated by no
specs I'm aware of.
I would, however, prefer a UTF8=ALL UTF8=FOO UTF8=BANANA style.
Dave.
--- End Message ---
