XSF Discussion - 2019-01-24

rion, I am aware of the email you are referring to; I have to think about it.

thanks for reminding me

rion, this has been requested several times, but I haven’t found a good way to implement it yet. Also, this needs a good way for the components to let the server know that their caps have changed. I have wire format for that in the pipeline, but I’m not really happy with it yet.

also, this is mostly server-side stuff then which I’m not at all familiar with

Is there a way for anything to say "my caps changed"?

for clients, yes

they re-send presence

for anything else, not

for _anything_

Zash, you’re a server dev. what would you prefer? send an IQ to all your open outbound streams? a nonza to all your open outbound streams? have a pubsub node on the server JID itself to which interested peers can subscribe?

(the latter is then like the first option, except that you need to keep track about who is interested)

What's the ultimate goal?

Zash which metadata ?

Zash, 1. provide clients with caps of JIDs related to their server (e.g. a MUC component) during connection startup (e.g. via a stream feature or something) 2. provide servers with means of learning their peers caps and keep them up-to-date (this was requested), and possibly broadcast changes of remote caps to interested local clients

edhelas: It's a dataform, you can make something up. I'd suggest something with an URI value. Then stick an pubsub URI/L or whatever there.

Zash this will require changes on the XMPP server, no ?

jonas’: for 1) it's probably fine for the server to disco#info external components as they connect and stick a caps-hash in the stream features next to its own (needs some origin indicator)

Zash, what do external components do when they change their caps for whatever reason?

jonas’: atm? nothing.

no, what are they supposed to do when that happens?

for example, change to an http upload component’s maximum file size

jonas’: I guess that's what we need to figure out.

exactly :)

So yeah, I guess we want the new caps hash pushed to whoever's interested, probably in some container, most likeley some way to enable it.

I do wonder if we need to ditch the current component protocol, since it's without stream features

Does it need to be global or local to the stream?

"it"?

jonas’: "caps push" or whatever

what does "global" mean then?

in the end, it should be able to propagate across domain boundaries I guess

jonas’: do you(r client) want to be subscribed to eg all the MUCs you ever heard of?

probably not

or, hm, your server maybe

but those one is joined in might be interesting ✏

jonas’: "global" as in something that uses stanzas, as opposed to a nonza-thing

I was thinking of using <message/> actually because messages with unknown payloads cause the least trouble

headlines even

yeah... for s2s (and component<->server) that might be good enough

and when done cleverly it can be implemented with a pubsub component. although that gives me flashbacks of Push

if it's something clients opt-in to then the payload container shouldn't matter

maybe some science is in order, like, what disco#info queries do clients send and to what

@board I may be on mobile today unfortunately Wish it worked ^

> send an IQ to all your open outbound streams? What would be wrong with a roster-push style IQ for changed caps?

Ge0rG, client2s or s2s or component2s?

everyone2everyone!!

wfm

jonas’: s2s and component2s

but yeah, having to disconnect and reconnect in the client is sucky. Also some clients will cache server caps in RAM to not play that dance every single time

on connecting, one could send the caps in stream features for s2s, too

jonas’: this doesn't even start to address the caps-change problem

Ge0rG, that’s true, but I don’t get your comment about reconnect

jonas’: the solution to update the client-side server-caps hash is to reconnect

jonas’: except when it's not

ah, I see

I don't think I made it more clear now.

enough to nudge my parser out of the local minimum and find a more sensible interpretation of this exchange

Why is everybody so unimpressed by the 0280 LC?

Again?

Zash: yeah. Apparently last year's council never completed the LC

But as the XEP hasn't changed in years, it still sucks and all my arguments from 2016 still are true.

quick poll: should I make XEP-0410 "Informational" instead of "Standards" before Proposing?

Is this legal?

Ge0rG, I think switching tracks is a rather undefined thing to do

Ohnoez. So now I need to ask Council for a decision on that, which will hand it over to Board for some serious XEP-0001 yak shaving.

410 is Standards.

Phew.

Wait, I thought there was going to me a disco feature in there, did I dream that?

Kev: Do you have a good explanation of why?

The only wire protocol it adds is the disco#info tag

that’s probably enough

Phew²

Zash: Because it introduces a new disco feature.

Or I thought it did, but it doesn't seem to.

Kev: https://op-co.de/tmp/xep-0410.html

I tend to agree with Kev; it essentially defines a behaviour for IQ ping on MUCs, with feature discovery.

that’s not just Informational IMO

Ge0rG: Ah.

Kev, it’s on the TODO, but Ge0rG didn’t want to make changes during LC

Kev: PR is https://github.com/xsf/xeps/pull/739 - I'm just preparing it for Council submission

Kev: and one of the LC notes was "Informational" or "Standards"

jonas’: now that the LC is formally over, I can ask you any time to merge the PR, right? (I'm not doing it yet, though)

I have to read up in '1 on that

the LC isn’t formally over until Council has voted, is it?

jonas’: until Council has voted about what exactly?

Advance vs. Reject

jonas’: and it will vote that on the XEP version submitted for LC? Regardless of two weeks worth of LC feedback?

Or will it Reject and I need to get you to apply the PR and then to re-LC?

sorry, I don’t have '1 in front of me

nm.

I've just sent a mail to standards@, and I'll kindly ask Dave to put the PR+XEP for a vote next week.

so I'd like your point of view on the experiment that I made to add avatar support in Pubsub, basically the idea is to publish 0084 items directly into the Pubsub node as items, this doesn't requires any changes server side, configuring the notifications without payload is adviced

the only case it's meh is when a getitems is done on the node, then the base64 of the avatar is delivered

on other cases, it's the same system as 0084

edhelas, ugly

you have a notification headline of the urn:xmpp:avatar:metadata item, you request the content, check the id and request the rest

defeats the entire design of '84 with the separate nodes to handle the issue with the large payload

will not work properly with node item count limits

mixes data and metadata in the same "stream"

ok :)

jonas’ what is the issue of mixing different kind of content in the same stream ? we have namespace for that

on my side it's the only solution that I found that doesn't requires big changes server side

and I'll not wait for MIX to have such basic features available

you could have a node with items for all other nodes

you could have an URI/URL in node metadata

this requires to set rights on the other node to publish content and restrict them

only the publishers can publish the avatar

and url/uri is possible yes, still in the mood of 0084 afaik

edhelas, as you mentioned yourself: "the only case it's meh is when a getitems is done on the node, then the base64 of the avatar is delivered"

item counts will be off, too

you have to do additional filtering, interactions with RSM and so on

on my implementation, displaying only 9 items and not 10 on one page is "ok"

but indeed it's not perfect

but RSM with chat message correction can also gives you 28 messages for 30 requested in the end

"Seve> @board" < I'm happy it doesn't. That's not something that should be in <body/>

pep., I don't disagree, but give the feature

:D

It should be in body, and then there should be a reference to annotate it.

The "@" is metadata, and is not required in body

You can use it as UI, but then not send it. Like Converse.js

yep

Although converse still fails in some cases, but that's the spirit

Zash I'll try out the URL in metadata thing and not publish the base64

URI?

https://xmpp.org/extensions/xep-0060.html#impl-uri ?

no, example 4 in https://xmpp.org/extensions/xep-0084.html#process-pubmeta

Why http?

where do you want to publish it ?

on another pubsub service ? which one ? the same as where the node is ?

we don't have any kind of filtering feature when ding disco#items on pubsub service, that's also why I moved comments nodes

it is possible but having simple HTTP URL is also fine, this node will only be about pulling data

(in my case Movim can even host it)

So we use http "because my client does it anyway" and it's a really ugly hack just because nobody is pushing for "the right thing"(tm)?

0084 allows it, also base64 handling through xml is having limits

i don't have control over the upload process of the picture for example

if I could take one day off to attend the summit remotely, which should I take? 31st or 1st?

Hard to say. I think often the first day discussions are more intense than the second as we prioritise the stuff more people consider important/interesting.

hard trade-off

thursday is burger day at work. I’d be missing good burgers.

>30 participants on the list. Not bad.

Looks very promising this year

And I have no way to get out of this project. I hope for a calm week so I can at least listen in

XSF Board Meeting | Logs: http://logs.xmpp.org/xsf/ | Agenda https://trello.com/b/Dn6IQOu0/board-meetings

sc/me bangs gavel

0. Welcome + Agenda

Hi. Who? What?

Here on the phone, though

Careful to not break it.

Haha

o/

Good one

Mattj sent his apologies

hi

1. Minute taker

Who?

Nyco has a different nickname? (or maybe it renders differently, with this update I had for my client). As we're not verified by MUC membership, anyone could pop in here now under the name 'Mattj' and vote in his name, right?

sorry, got a backup client, my default one had the bookmark removed, will investigate

Guus means: start implementing MIX.

Doesn't matter Nyco - Just wondering out loud

In any case, it uses the same real JID, so we're good.

No takers (haha!)?

2. Commitments

Call for Editors. jonas’ ?

I've seen the call, but no responses, iirc

Shall I remove the item?

I'm not sure what more that we can do, other than keep doing calls

And since we're not actually doing them, I'm removing the item.

* Clarify how initiate last call for Exp. XEPs

that was in your PR, right?

I put in a Pull Request for this, along with how to actually propose.

https://github.com/xsf/xeps/pull/744

It would be nice to get reviews on this.

I've read and remember agreeing with it - I'll go over it once more after the meeting and +1 on the PR

do we need an in-meeting vote on this?

I put it up for review

Don't know what MattJ's handle is

on github it's @mwild1

mwild1

He doesn't show up as an option, so I'll look into that.

It might also be nice to get opinions from the floor

and council/editors, specifically. They work with this stuff

Right

* Update Board status on members page

This sounds like an 'Alex' item.

ralphm: might be useful to bring the PR up for Council then. I kind of missed it

(I've just had a scan of that PR and have issue with it, FWIW, in places)

Ge0rG: the point is that Board is the Approving Body.

Kev: great, all feedback is welcome

Board is the approving body, but I think it'd be irresponsible to make significant changes to the standards process without consulting the body responsible for the standards :)

I don't think there's a rush for this, so I suggest getting it put on Council's agenda too.

https://xmpp.org/about/xsf/members <-- has board members?

Kev: I'm happy for it to be on the Council's agenda.

Ta.

Guus: right, I'll archive it

I'm unsure if I read the meaning of "Update Board status on members page" right.

I thought we were talking about the update I did not that long ago

Kev: for what it is worth, I had already mentioned this PR on the standards list a week ago.

ralphm: I think it's worth flagging things explicitly on the Council agenda.

Kev: that was the goal already and why we are discussing it. The Council's Chair is invited to Board meetings explicitly for these things.

(and also on the Board mailing list)

Anyway, I trust it will be discussed in the next Council meeting now. Meanwhile comments are also welcome on the PR. Moving on.

3. GSoC

flow: updates?

there's been an announcement, blogpost, wiki page, and revival of the GSoC that's since silent.

right

https://wiki.xmpp.org/web/GSoC/2019

no proposals as of yet, it seems

The wiki page still has the default templates indeed

4. XMPP Summit / FOSDEM

Updates: van arranged, 500x keychain bottle opener ordered, 100x hoodies ordered.

dinner invite went out

I sent out a reminder regarding the removal of vending machines at FOSDEM.

the hotel signup thingy has expired, so if you're still looking for a place, you'll have to make your own arrangements. You can probably still use Thon EU if you want.

So far only 16 responses for Dinner. Another reminder to please fill in the form. No response, no food!

I'm happy with the turnup at the summit. It's more than last year.

17

people should write down suggestions for agenda items on the wiki

who's this "Allergies: everything but filet mignon" <--??

31 registered people for Summit

Kev of course

It was suggested to me, but I didn't.

moving on?

I can't stay to long, today

did we loose ralphm ?

no

yey

5. JabberSpam trademark application

JabberSPAM actually

Ge0rG's response to Peter, did that get a response?

Guus: not from Peter, AFAICT

No

The project v.s. organization thing is a bit of a thing.

I can see how having another organization with Jabber in its name might complicate the trademark situation.

And hence why Peter said that might not be a good idea.

is "that might not be a good idea" equivalent to the board denying the request?

I'm a bit lost here about how to go on.

there's precedent for having an org with 'jabber' in it's name, trademarked: https://xmpp.org/about/xsf/jabber-trademark/approved-applications.html

I wonder if they paid the 500$ fee.

I actually heard once that the fee was never imposed.

I can only assume it was, but don't know.

as on how to go on: I'd like to get Peters feedback, as he's most knowledgable on this

Well, as Ge0rG said, it might not fit the description of project either. I am struggling with this myself, and have started reading all the documentation linked from https://xmpp.org/about/xsf/jabber-trademark/background.html

from a process point of view, it would be great to have this process terminate eventually with an official board response.

I've read that - there's nothing conclusive in there that I could find.

best I could reason is that George is putting forward two projects, sharing (part of) a name.

(the manifesto and the blacklist)

Guus: this is a question similar to "does a github organization count as an organization in terms of the trademark licensing agreement"

Ge0rG you're right.

https://xmpp.org/about/xsf/jabber-trademark/license-decision-process.html lists the options

As long as there's no legal representation of the github org ...

a github organization is not a legal entity, I think.

It also says that the XSF Trademark Committee makes this decision.

the XSF Trademark Committee is Board + Peter?

It is comprised of: “a XSF Director and XSF Members who have been authorized by the XSF Board of Directors to manage the trademark licensing program”

The former we don't have

And the latter we don't have either

and the latter we may not have explicitly appointed

so it's up to board to pick up the slack there.

I think so to.

too

let's not complicate things furhter.

If I volunteer to the XSF Trademark Committee, will I be put into a position to decide upon my own request?

I tend to have opinions other than Guus and mine.

(on anything in this meeting, really)

want to have, I meant

you have other opinions that you? 🙂

ah 😃

FWIW, I tend to approve, if we get the details correct.

Could that be changed for just Board for example? So we do not have to go though that process

I would assume that the XSF Trademark Committee implicitly is the XSF Board.

Seve: I suppose Board could appoint Board to be on the Committee.

Seve, I don't think we have to go to that process to begin with.

But I think Board is clearly authorized if it votes on the matter.

I am fully OK with the Board voting on trademark applications.

Also deciding on whether to ask for one of the fees.

(personally, I'd love to change the process to not demand any fees from non-commercial entities)

I'm not considering any changes to the process before understanding how it is currently supposed to work.

Ge0rG working on the assumption that Peter does not bring up new issues, you change the name on the trademark application to the non-abbreviated name (as you already suggested), and we're licensing to two projects, instead of an organization, I tend to be OK with issuing the trademark. Let's ping Peter, and find out more.

Guus: I'm not sure whether my mail counts as "pinging Peter"

I'll ask Peter for an update

Guus: also one of the open questions was whether the long-name trademark will allow me to use the shortname in URLs

Right

Thanks Guus

and that question is kind of blocking for me

I understand.

because if the answer is "no", the whole application stops making much sense.

I assume it is OK, but would like Peter's feedback on that.

Let's try and get more answers for/during next week at the Summit.

6. Executive Director (bis)

thanks!

I know we have discussed this at length already, but I've been reading our bylaws, and I'm not too comfortable about the vacancy for this officer.

what are your concerns?

The point is that the bylaws (and e.g. the trademark policy) explicitly mention this role.

And it is unclear how things should work in the absence of an ED.

from the floor, I'd assume that All Of Board needs to make any decisions then.

Right, but I wonder if that should be made explicit. Unfortunately, we didn't really get much feedback on our previous meeting on this topic, either.

The minutes now state: “As of now, Board is executing their decisions themselves. The consensus seems to be that this mode of operation is fine for now, and there is no extreme pressure to find a replacement.”

I think that is aligned with Ge0rG's comment.

I'm not to worried about not filling the position, and not change the bylaws either.

Ok

if a problem does occur, we'll have to resort to appointing another ED.

which will then be problematic, probably.

but in effect, the ED has been absent for years.

and/or done nothing in that role

What problems could be encounter because of that?

it's way past time for me

I cannot currently think of one. So maybe we should leave at this for now.

7. AOB

Ok, no problem.

I'd like to vote on the ED thing asap

it's been dragging

AOB: nyco, did you get in touch with that designer friend for the badges?

Guus: what exactly would you like to vote on?

Ge0rG to keep the position, but vacant.

Ah, yeah. The badges. Would be great to have them.

Guus: I'll put the vote on for next meeting.

nyco ?

Ge0rG: Postponing that item.

8. Date of Next

lets do +2w

Next week is the Summit. We may meet in person, but I'm not scheduling a meeting.

Suggesting +2W

(also better for Minutes and following in here)

Ge0rG: any formal meeting would also be in here

Unfortunately I will not attend Summit nor Fosdem this year, so I'm okay with what you suggest

(transcribed or otherwise)

9. Close

Thanks all!

Thanks, bye!

Thanks Board!

thx

Thank you guys

XSF Discussion | Logs: http://logs.xmpp.org/xsf/ | Agenda https://trello.com/b/Dn6IQOu0/board-meetings

> Stream closed by us: Improper 'from' attribute (invalid-from) That had to hurt.

Is that even legal (assuming the server did it in response to the clients 'from')? ✏

moparisthebest: hi...

hi

Andrew Nenakhov, it seems Xabber asks for the vCard-temp of the contacts on their full JID, and twice in a few ms, when the contact connects.

Link Mauve: https://github.com/redsolution/xabber-android/issues/839

Oh, thanks.

It is doing it twice though, which isn’t described in this bug report.

Link Mauve: once is one too many already, but please add a comment

Ge0rG, done.

Link Mauve, tried a XEP-0030 request? some software includes a version there in some way

jonas’, that did it, thanks; I just edited the comment.

:)

Also, since you were all waiting for it, poezio now supports SVG avatars. o/ ✏

does it raster them to a pixel graphic before rendering them as character art first?

Of course.

To a 16×16 buffer.

The character art is just either Unicode half-blocks, or full-blocks, depending on whether your ncurses is broken.

boooring ;-)

should do fancy tricks with \ and / to approximate lines ;-)

Now get me 2x2 pixel avatars next to nicks in rooms

jonas’, we have Unicode support, there is no reason not to use it. :p

Zash, the code to do that is here.

I’m not that sure about the usefulness of it yet.

Something something human visual identification of symbols and colors and something

I think it might actually be useful

I probably should add a cache of the rendered pixel data first though.

https://upload.bouah.net/upload/4q151Oj_FHU3D1v9/poezio-avatars.png

Otherwise rendering the participants list might become slow.

For those who wonder what Link Mauve's avatar looks like

▗▚▜

pep., cute isn’t it? ^^

Zash, jonas’, if you find a nice algorithm rendering finer pixels, I’m all for it.

Maybe someday we'll implement MAM in poezio

This looks a bit like motion detection in video codecs.

Link Mauve: U+2596 .. U+259F

When we're done with avatars

Zash, the issue is about identifying which two colours and which block character to use.

Those + for- and background colors gets you 2x2 pixels in one square

you could copy whatever gitea does for generating avatars

Link Mauve, with SVG, you have the additional data

you don’t need to do much guess on the pixel data

but I find them basically indistinguishable after more than 2 or 3

jonas’, not with this renderer though.

Link Mauve, write your own!

And we still need to quantize to two colours per actual character.

jonas’, no, I won’t write a new SVG renderer. :p

aww

Link Mauve: so you are back at fixing poezio, after fixing the prosody caching module? ☺️

Ge0rG, I didn’t manage to reach anything satisfying there yet.

https://cerdale.zash.se/upload/Hbr_S1X3EHlZZ_RR/prosody.png

Yes, these two characters are perfect to represent Prosody’s logo.

What more could you possibly need? :)

I’m pretty sure most avatars in this MUC won’t render that well. :p

That's one character tho. (followed by a space because my terminal is weird)

which one is it?

st

the character

It doest't like double-wide chars

not the terminal…

I wanna see how it renders in mine

jonas’, ▚.

▞ U+259E QUADRANT UPPER RIGHT AND LOWER LEFT

There's a few around that one useful for drawing pixel art