Re: [hybi] WebSocket feedback

[Greg Wilkins <[e-mail]>]

Joe Hildebrand wrote:

> Keep in mind that we'll need to have consensus over the whole document by
> the time we're done, so I don't see a need to propose individual chunks of
> text before they're inserted in the document, particularly at this early
> stage.

Really?

I find creating a consensus for inclusion on merit far more
positive and productive than trying to find a consensus about
exclusion due to problems.

Otherwise, it makes for a bit of an asymmetric relationship
with the editors.

If I make a proposal, I've got to convince others that its a good
idea and then it should get included.

If the editor(s) have a proposal, they can just put it in the
document and if it is not liked, we have to have a create a
consensus to have it removed.

Take this latest idea of including a random string in the
stream after a valid request/response which goes significantly
beyond  anything that was discussed on the list.   Surely a
proposal like that should have raised on the lists first and
only if some support for it was achieve should it go into
the draft.

As it is, we have a solution in the draft for a problem
that we've not yet even fully defined or discussed.
Surely we should strive to work from requirements towards
solutions rather than the otherway around?

Note only that, it makes it hard for implementations that wish
to track the draft.  ie should I be updating my imple now to
match the draft, only to have to revert later if the idea
get's knocked back?



regards