XSF Discussion - 2019-07-01

mam

Am I right to conclude that adding support for XEP-0157 as well as XEP-0232 on a server will result in more than one service discovery extension to be added to the disco#info response (which, according to https://xmpp.org/extensions/xep-0128.html#impl is "unlikely")?

yes

and i think i've seen clients do that

because it triggered a bug in my caps calculation

(besides the point: what two extensions are added by clients? XEP-0157 seems applicable to servers only?)

i was wondering that myself. but i don’t recall. that bug was fixed like 4 years ago or so

What 128 usages are there besides those two?

don't know. But these two now seem to bite the naive implementation in Openfire that defines that this shall not happen. 😃

Hm, there's the http upload thing

Altho also mostly for servers

Not that you couldn't do p2p http upload if you really wanted

or yeah maybe it was prosody

for when prosody does http upload on the server domain

In prosody it's handled through the same internal method (roughly named lists of items attached to the host, tracked per module), so it shouldn't have any problem with arbitrary number of forms. Should be no different from features.

sure. i just meant maybe it was prosody (and not a client) that triggered the multi form bug caps calculation in Conversations

right, makes sense

because with http upload and 157 you end up with two forms on the server domain

thanks!

and servers have caps

should the remark in https://xmpp.org/extensions/xep-0128.html#impl be modified?

seems less unlikely to happen.

on one hand just remove that sentence seems reasonable on the other hand it's more than just editorial and we probably can’t do that to an active xep?

but if we can i'd vote for removing it

Could one of you draft a PR?

then it can be discussed in this weeks council meeting

I’d like to note that I find it curious that a XEP which defines wire format is only Informational :)

Perhaps it's actually defined in XEP-0030?

i mean the wire format is 0004

Hm, no mention

and 128 is probably meant as. 'here is an idea why not use 0004'

Why not put 0004 everywhere!!!

rebase all of XMPP on top of 0004!!k

0004 in pubsub!

jonas’ https://github.com/xsf/xeps/pull/797

Guus, \o/

is PAM still a thing?

Sorta needed by MIX, no?

mhhh i was thinking about regular PAM not MIX-PAM

"In future, this specification MAY be incorporated into Pubsub Account Management (XEP-0376) [4] (PAM) which follows a similar model. " <- maybe we should make that decision before we roll out mix

and not in a future after mix was deployed

> Zash> Sorta needed by MIX, no? unfortunately not

Because we concluded that you could do presence based MIX?

no MIX does simply not re-use xep376 semantics

or, it does re-use the semantics but not the protocol

Imho it would be kinda neat if it did. Because then we would get regular pam for free and can use that in other places as well

Hehe, funny that XEP-0392: Consistent Color Generation already exists in two incompatible versions :D

Daniel, totally aggree

vanitasvitae, which ones?

YCbCr and HSLuv?

https://xmpp.org/extensions/xep-0392.html#revision-history-v0.5

And CRC32?

So it doesn't exist in two incompatible versions, it exists in HSLuv, no? :)

(I know dino uses a slightly different one, but they're similar enough that jonas was going to add it to the XEP iirc)

http://www.hsluv.org/comparison/ I think dino uses HPLuv

If I'm not mistaken, version 0.4.0 produces different results than 0.6.0 does

vanitasvitae, I guess the point is to use the latest version :)

Yeah but still :D

Things evolve

That's what happens to Experimental XEPs

To anything! To Life!

It also happens for Final XEPs, just not in a direct way. A new XEP gets created and people are encouraged to move to it, and after a few years this new XEP is now the "modern" way of doing it

Stability is an illusion

Is this turning into a Debian rant? 😛

To be clear, I just find it funny :D

I wasn't going to mention it :p

Zash, though, Debian's definition of stable is different

i should probably give hpl a try for Conversations as well

vanitasvitae, there’s a reason I haven’t asked for a Last Call yet ;)

jonas’: can the authors decide for themselves when to issue the last call?

The author has control over the experimental XEP and decide when to LC yes

They can ask council

Ah ok

anyone can ask Council

vanitasvitae, and you probably didn’t even spot the subtle differences for MUCs betwene the versions ;)

jonas’: jids vs. nicks?

yes ✏