[Standards] Council on Stanza Repeaters without Multicast
dave at cridland.net
Thu Apr 3 08:47:42 UTC 2008
On Thu Apr 3 00:56:51 2008, Tobias Markmann wrote:
> That's nice to hear for OpenSSL but there is GnuTLS, NSS(Mozilla),
> SSAPI(SChannel) and YASSL.
I can't speak for any of them, but my understanding is that
compression is pretty commonplace on "desktop" TLS stacks, now, hence
a very long and protracted argument about why it's not prevelant on
Certainly GnuTLS does DEFLATE according to RFC 3749. YASSL notes
"zlib compression support" on its front page. On a quick skim,
neither NSS nor SChannel docs were very forthcoming, and I don't have
the data off-hand, but it'd surprise me if they don't as well.
> I doubt compression can be used between any two
> different implementations which is a pretty bad situation one might
> not want
> to base on.
No, that's tosh, sorry. RFC 3749 dictates how DEFLATE is used as a
standard TLS compression codec, and it's known to be interoperable.
(DEFLATE itself isn't known to be interoperable, oddly, but everyone
uses zlib anyway, and its usage within TLS is known to work.)
> Though it is nice if servers can use TLS' capabilities of
> compression it seems better to me also or only implement usage of
> before TLS.
Both is fine. OpenSSL, at least, allows applications to check whether
compression is in effect, allowing (in this case) servers to decide
whether to advertize, or attempt, XEP-0138. But lack of compression
support in TLS should be considered a bug at this point, and I see no
reason at all to avoid using it. (I'm not sure how to, without simply
using SSL instead of TLS.)
Dave Cridland - mailto:dave at cridland.net - xmpp:dwd at jabber.org
Infotrope Polymer - ACAP, IMAP, ESMTP, and Lemonade
More information about the Standards