Hi all,
A few of us just had a quick chat about some XEPs we think need to be
modernized. We came to a rough consensus on what we thought would happen, and
they're mostly in line with what's been discussed on list. Here are the notes
we took:
Present:
- Sam
- MattJ
- moparisthebest
* XEP-0138: Stream Compression, XEP-0229: Stream Compression with LZW
* MattJ: everyone is streaming video on mobile devices so XML
is not a big
resource hog that we need to worry about anymore
* Sam: for large deployments compression can actually save CPU
cycles
(fewer TLS packets to encrypt)
* Moparisthebest: thinks someone (Zash?) argues that a new
thing is needed
with better negotiation to avoid the security issues
* **Recommendation:** deprecate and replace, there are newer
algorithms
that will need different negotiation (zstd with fixed dictionary?)
* XEP-0114: Jabber Component Protocol, XEP-0225: Component Connections
* MattJ: it's been a decade, no progress, no real benefit to
moving from
0114 but the new one does SASL and TLS
* Sam: SASL would be nice, TLS seems like it could just be done
directly or
w/o the component and servers knowledge and be fine
* Sam: structured my XMPP library differently because I knew
I'd want
multiple different types of stream handshake, maybe it
would have been
simpler w/o knowing 0114 would be coming in the future
* MattJ: chicken-and-egg problem. Not enough gained when
switching but we
just need implementations in prosody and ejabberd and
the ecosystem will
follow
* Sam: is namespace delegation easier with 0014 or 0225 or
completely
orthogonal?
* moparisthebest: if a new person implements this how do they
know which to
do?
* MattJ: One has green text, one has red text.
* MattJ: which is easier if you're implementing
from scratch? 0114
easier from scracth, 0225 easier if you
have an existing XMPP
handshake library to use
* **Recommendation:** not high priority, no action needed for now.
* XEP-0072: SOAP Over XMPP
* **Recommendation:** Obsolete
* XEP-0054: vcard-temp, XEP-0153: vCard-Based Avatars, XEP-0292: vCard4 Over
XMPP
* moparisthebest: make large stanxa sizes
* MattJ: trying to move away from vcard-temp in Prosody, not
much interest
from anyone else (Gajim maybe has newer stuff),
official support
deprecated in next Prosody (support provided through a
backwards
compatibility module)
* MattJ: privacy controls are in the new version, that would be
a benefit
to moving but does not create a compelling need to move
forward, but it's
one less thing to worry about (everything is PEP)
* MattJ: one weird thing from a spec perspective is that it
defines an IQ
to fetch/set vcard, but it's PEP based so you can just
use PEP. Maybe get
rid of that?
* **Recommendation:**
* start LC on vcard4,
* bring up why it has an IQ and a PEP access method,
* hopefully move vcard4 forward
* put it in compliance suites (2022?)
* bind2?
* Carbons/MAM race condition
* Persistent device tracking
* MattJ: bind2 needs new author or someone with more
time then it should
be fairly easy
* everyone has to leave around this point
Thanks for participating, and we hope the council will consider some of these
recommendations if they're not already. There were only three of us, so more
discussion on list would be great!
—Sam
--
Sam Whited
_______________________________________________
Standards mailing list
Info: https://mail.jabber.org/mailman/listinfo/standards
Unsubscribe: [email protected]
_______________________________________________
