Re: [hybi] Payload only compression extension, again

[Brian <[e-mail]>]

FYI, the WebSocket API draft has already been updated to support binary
messages.

Brian

On Mon, Jul 25, 2011 at 4:56 PM, Greg Wilkins <[e-mail]> wrote:

> On 26 July 2011 01:05, John Tamplin <[e-mail]> wrote:
> > That would also be useful if there were heuristics to choose when to
> > compress the data.
>
> Once there is a RSV bit used, then the sender can use a variety of
> different strategies to determine if to compress or not.  Heuristics
> is a good one and I think most server gzip filters apply simple
> heuristics (eg if uncompressed is > size and if compression does not
> overflow buffer of same size).   Although servers also have the
> benefit of mime types (eg don't try to compress image/jpg).
>
> > However, as long as the compression state is maintained across frames, I
> > don't think it is a real problem to simply compress everything.
>
> I think that is the simplest and best heuristic to start with, and
> will probably be beneficial in the vast majority of cases.
>
> From the browser side of things, it may be the only practical
> heuristic while the API is just simple text messages.   Perhaps when
> binary messages are supported something else might do better, but
> that's the benefit of using a RSV bit, as the strategy can evolve over
> time without the need to update the protocol specification.
>
> cheers
> _______________________________________________
> hybi mailing list
> [e-mail]
> https://www.ietf.org/mailman/listinfo/hybi
>