XSF Discussion - 2017-11-22

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

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

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

Congratulations to the new council members!

erm... And the new board members, too. :)

(Somehow council elections were more prominent to me.)

dwd, Kev: yes I'm pretty mich available all day. So 1630Z or what ever you suggested works for me

Fab, ta.

I'm very hopeful regarding 1630Z as well. Might be a bit unresponsive while entering a train, but not absolutely unavailable.

I've added the Minutes link to https://wiki.xmpp.org/web/Board_and_Council_Elections_2017 and now I hope I'm not breaking the page-copy-paste process for next year's election.

Who's formally going to apply the tie-breaking process for the remaining Board position? The secretary? Old Board? New Board?

Guus: was quickly reading RFC 3797 last night, but havent really understood how this algo works, needs more reading ;-)

Alex: TL;DR: we need a pool of candidates (check) and an algorithm that will be applied on a future random-generating event (e.g. state lottery numbers)

Alex, Pick something happening soonish (like, say, Euromillions draw). Define how we get from that to an answer. We could just hash a string of the lottery result as 2-digit decimal ASCII numbers in ascending order. Then pull the least significant bit. Odd is Nÿco, even is Arc.

dwd: should the ASCII string be terminated by a newline?

Ge0rG, I don't think it makes any (predictable) difference to the outcome, so we may as well for simplicity. I'd pick a strong hash algorithm just so we can claim that anyone fixing the Euromillions draw purely to pick one or other wouldn't know what to fix it to.

Although if they can fix the draw, please let me know so I can buy a ticket for once.

the RFC talks about combining various sources (eg: Euromillioins and the value of some stock exchange by the closing on a particular date in the future)

dwd: let's just add a foot note that in case an XSF member wins a significant amount of money, we need to declare the result as void and choose another state-run lottery.

dwd, :D

Guus, Yes, but that's to increase the entropy required, which is given in a formula - we only need a single bit as I understand things.

dwd, agreed. And as our pool size is fixed to '2', a simple odd/even determinator will work.

I'm primarily interested in making sure that someone is actively persuing a resolution at this point though :)

RFC3739 §4 provides a string normalization algorithm.

I am not a gambler on only familiar with the german lottery, every weekend they pick 7 numbers out of a pool from 1 - 47 if I remember correctly

i thought 6?

ah, maybe seven, but the seventh has special meaning or something

jonasw: jonasw no idea, could be also 6 ;-)

we could use the least significant number in a stock exchange course?

Alex, not to add to the workload, but perhaps you should put up memberbot for the membership application q4?

submission deadline was yesterday

Guus: yes, will be available later toda

tx

memberbot is up for the next voting period ;-)

Looks like we have some Non-Reappliers?

thanks Alex

Ge0rG: yes

I usually push them and remind several times by Email

Right, the point of no return has been reached. There will be a Q5.

(Y)

I've scribbled a suggestion for a concrete RFC 3739 implementation for us on the members@ list. It's dated for Friday, which seemed like a reasonable timeline. If there's something concretely wrong, please point it out. If you're happy with it, please say so. And please don't turn this into a bike shed.

why not just take the last number of the euro million draw, and say if its 1-6 candidate 1 wins, 7-12 candidate 2

i dont get why there is hashing needed

dwd: you take the last byte E0 in your example, or just the last char 0?

or do we just go from ricght to left until we find a "0" or "1" char?

Alex, Bit - 0 ends in a 0 bit. But loosely, odd numbers Nicolas, even numbers Arc.

lovetox, Makes it closer to RFC 3797, basically. It's certainly the most over-engineered coin-toss one could possibly wish for, I know.

Alex, So last characters 0, 2, 4, 6, 8, A, C, E would end in a 0-bit. The rest in a 1-bit.

ya

just did the Friday draw

data: 20.26.35.36.42./ hash: a33d86c56740f1291fde14dd6b89b371644a9023834ff71665bd79160dd90719

ending with a 1 bit, so would be Nicolas

Alex, I get the same result, indeed.

I second that proposal ;-)

was confused because you mentioned modulo in your email

Alex, Yeah... If you treat the hash as an enormous integer and modulo by the size of the list, it's the same thing. Just expressed in terms of the size of the list - which we know to be 2, so it's simple.

its true that you cant predict what the result on friday may be, but did you make sure both candidates have equal chance of winning?

these are two different matters

dwd: ya, next time when we habe a tie with 3 or more ;-)

Thanks Dave

looks fine

It'd be good to get agreement from arc and nyco, mind.

lovetox, We assume that the lottery is completely random across its range, and that the hash output has a uniform distribution of bits. I believe the first is a reasonable assumption, and I believe that disproving the second would lead to a sudden move to SHA-3.

As long as it is independently verifiable

So, link to this lottery ?

Zash, In my email - but also https://www.euro-millions.com/results

And a clear description of the algorithm so anyone an run it

Zash, Also in my email.

"That is covered in my email" ?

Zash, https://mail.jabber.org/pipermail/members/2017-November/008696.html <-- That email.

Thanks for picking this up dwd

ralphm, No prob. I felt it'd be better coming from someone off Board anyway. And I dimly recall being partially responsible for the RFC 3797 rule anyway after last time. (Which, as I recall, also involved Nyco - though I genuinely can't remember).

ralphm, Ah, no, I recall slightly wrong - the tie was between David Banes and Michael Rémond, but Nÿco was the "fourth" member, and we voted to drop the Board to the top 3, not 4, in order to maintain an odd number - which lost Nÿco. https://mail.jabber.org/pipermail/members/2009-October/005230.html

dwd: you forgot to explicitly specify the ordering of the lottery results.

Ge0rG, Isn't this enough? "2) Each number will be arranged in ascending order, separated by dots, and terminated by "./". "

dwd: oops, my bad, sorry!

Ge0rG, Don't worry. I'd rather people pointed things out and were wrong than an error wasn't spotted until too late.

dwd: still, reading the mail on my smartphone in a moving tram and immediately accusing you of negligence was impolite at best.

So my appointment ended early, which means I have the time for the council meeting. Now I just need to hope for good network coverage...

Ge0rG, We're aiming for 16:30 UTC, so 17:30 CET I believe. So you've more than an hour to find a café with WiFi.

Ge0rG, Also, you weren't impolite, and nor did you accuse me of negligence. You reviewed something important and thought you'd spotted something, which is very welcome.

dwd: I'm boarding a train in 10 minutes, and it will have bad wifi. I can't influence the network quality except by pulling the emergency brake, but that will considerably delay my arrival at home 🤔

Ge0rG, We'll have to hope we did a good job with all those mobile extensions, then. :-)

dwd: the largest blocker on the first trip today was poezio's broken reconnect code and that nasty MUC self ping IQ routing problem

Ge0rG, It'd be nice to get MIX moving again.

it would be nice to burn it with fire ;)

Indeed, I've recently made another approach at the XEP, and I've read almost half of it. Unfortunately, I wrote my notes as messages-to-self and they weren't logged because of infrastructure problems. They are now captured in the yaxim SQLite storage of a phone with empty battery.

dwd: in what way??

Steve Kille: make it a draft standard? :)

I think it's a little bit raw for that.

I was referring to the intended direction.

We want this direction

Need some implementations

sorry, my semi-funny remark wasn't helpful.

As for the Board meeting. I look like this right now, and using any screen terribly hard:

https://upload.ik.nu/upload/eb78d522-cde8-401f-acfe-045ede3645de/1YWY6o4HQRCx44MBO8aXHw.jpg

ralphm, what happened? are you ok?

Damaged my cornea somehow. Should be fine in a few days

ouch. Get well soon, Ralph

ralphm: get well! and don't overwork your other eye

Arrr

Welcome Guus especially

Who do we have

wait, you're an hour early?

Council hasn't met yet

Ah yes

Adjourned then

I missed the change of time

(Also, in an hour this building is closing down, so at best, I'll join the meeting via mobile from a parking lot somehwere)

May I suggest to Board that having a meeting before Board are selected is probably suboptimal, given the first order needs to be electing a Chair?

, .,

Oops Sorry

Well, hah, so first of all, there are 4 definitely elected board members and they can meet as usual. Second, I thought it would be important to affirm Dave's suggestion to resolve the tie.

There is no strict rule for the first meeting to elect its chair.

There isn't, but it seems odd not to.

Bylaws say one year or until elected. I agree all is weird. But the method to resolve the tie has to be affirmed in some official capacity, no?

No.

At least, I don't believe so. I think we fixed this in the bylaws last time it happened.

Kev: agreed. I think it's primarily up to Alex, rather than board.

> in the case of a tie for the final remaining position, the final individual shall be chosen in accordance with the procedures defined in “RFC 3797: Publicly Verifiable Nominations Committee (NomCom) Random Selection” published by the Internet Engineering Task Force. There is nothing about who is responsible for that process.

Kev, Yeah, the RFC 3797 mechanism is in the bylaws. The precise choices are not. The only affirmation I'd really like to see (beyond a few members giving it the nod) is for Alex (as Secretary), and Arc and Nÿco (as those affected).

Ge0rG: in all matters unspecified, I believe the board or the executive committee can do this

but is there a board until the tie is resolved? what if all the board election is a tie?

And I agree with dwd otherwise

There is currently a 4 person Board

This does seem like a thing done by those administrating The Meeting

moparisthebest, The former is unspecified. I think technically, we should pick the RFC 3797 process prior to the election. If the entire Board is a tie one year, the selection still operates by RFC 3797.

yea but not if the rule is 'the board has to agree to the algorithm'

ralphm, I'm not entirely sure if there *is* a Board. There *is* a Chair, though.

moparisthebest, Good point, well made.

Clarification of the bylaws in order? :)

Zash, Maybe. We could also declare that in the event of multiple interpretations of the rules, the Secretary's word is final.

Who has the authority to determine whether a certain process is RFC3797 compliant?

dwd: Sure. And then put that into the bylaws.

I think assuming lottery is random and not-riggable is fine, if you could rig it, why would you use that power to influence XSF board/council members?

I think that adding the Secretary as the authority sounds reasonable, but what happens if somebody vetoes the RFC3797 compliance?

dwd: I read 3.13 that all but the tied candidates for the last remaining position are simply elected

moparisthebest, Says exactly that in the RFC, actually. :-)

moparisthebest: "<Ge0rG> dwd: let's just add a foot note that in case an XSF member wins a significant amount of money, we need to declare the result as void and choose another state-run lottery."

ralphm, I can go along with that, actually. Particularly as it talks about "Election of individuals", suggesting they're elected individually rather than as a Board entire.

I'm not sure how much money would buy my board/council vote, but certainly a small % of a lottery win would do it :P

And indeed, I'm still chair. I'm not sure if stpeter is still executive officer, though, given the one year term and his non-appointment last year.

(6.2)

Does the new board start directly after the election?

Zash: Yeah.

Actually, I am wrong. He was appointed, and indicated intent to resign

Ok, what now then?

*shrug*

I suggested just just hold a quick meeting affirming dwd's method

Works for me

And then arrange a new meeting after Friday's lottery

IMHO, what dwd said is reasonable.

I think nyco and arc agreeing to it would be optimal.

Rights

-s

And also that in future elections, we should probably state this method in advance :)

Alex are you here, too?

Kev: it might be hard to find an appropriate wording for the bylaws that doesn't open up to potential abuse.

Good luck cheating the lottery. If you manage that, by all means be a Board member

Ge0rG: I don't think it needs going in the bylaws - the bylaws say what to do.

^ what ralphm said.

Kev: but not how to do it

Stating it in th emeeting announcement should suffice.

As far as I'm concerned, Alex simply stating that this is the mechanism is sufficient.

I agree with Kev

Anyway

Kev: unless somebody points out that it violates RFC 3797.

0. Welcome and agenda

Who's here?

I am

kind of

I'm here

in process of getting kicked out of building.

2.5 isn't really quorate

No MattJ?

I'll respond eventually :)

Here

Sorry

Arc, nyco ?

Ok all elected in attendance

I have one agenda item: resolving the tie

That's up to Alex, isn't it?

1. Resolving the tie for 5th director

So, there has been a suggestion by dwd to resolve this according to the bylaws

Additionally, there's some unclarity how the foundation selects this as accepted

I motion that the Board affirms dwd's method, having seen no objections

+1

Of course comments of Members are welcome

Although I'm perfectly fine with the suggested method, it's purpose is to safeguard Alex from accusations of bias. Its for him to agree on, I believe.

I consider dwd's proposal very reasonable, especially the part about approval by the two affected nominees.

I'd be happy to make it conditional on that

Sounds good to me with that condition.

I think that we actually more or less *have* to leave this to Alex to agree to.

Kev: can you explain why?

Because either the bylaws are ok for him to do so, in which case yay, or the bylaws need changing before we can have a new Board, and that means voting and blah.

To be clear, we *have* a new Board IMO

IMO it is compliant to RFC3739, but in case of incompliance there should be a reasonable vetoing approach. I think that having a single person who defines that it is compliant is insufficient.

Ge0rG: this is why I thought it useful to affirm by the Board. I like the suggestion of having the candidates needing to agree, and of course Alex will need to execute (and agree).

I was hoping all involved to be here

Well, everyone appears to agree with Dave's method

ralphm: right. What Kev said almost read to me like only Alex needs to affirm the method.

So perhaps Alex can just execute, assuming he agrees to the method too, and we don't have to worry about bother potential scenarios.

Agreed

Kev might be right that his affirmation is all that is required. However, that does not mean that board and member affirming that's the way to go isn't a good thing.

Especially if it's affirmed by all parties directly involved.

I'm fine with this

intosi: I think this boils down to who is allowed to veto

nyco?

I don't think we can sensibly do more now, with all other parties absent.

Ge0rG, Push comes to shove, the members can remove the whole of the Board and redo the election.

^ I tried to say what dwd said, but didn't manage to find words even closely as eloquent.

I think Board can also appoint the vacancy, so that's why I think it is sufficient for the Board to agree.

ralphm: that'd be weird. We could fill the vacancy to force a vote in a particular direction that way.

ralphm, No, not also. We need to complete this first. I'm not even sure if, technically, either of Arc or Nyco can withdraw.

(4.7)

I'm happy to debate this all day, but I think we have enough to go on

Let's put this on hold until Alex responds.

I agree on Daves suggestion and I am happy to execute on it

Honestly I think way more thought is being put into this than is needed. Let's see what the Friday lottery results in and move forward

Good.

Dave said the next draw in Friday, so we should have the 5th borad member by the end of this week using the random election method

Ok. If nyco does not object before the lottery executes, this is what we do.

dwd: can you write minutes?

2. Date of Next

tbd on list

What list?

ralphm, Yes, I'll knock something brief out.

3. Close

Guus: board list

I'm not on board list yet, I believe.

Guus, board@, one assumes.

Kev can add you

Thanks all!

Thanks

Arrrr

Cheerio all

Roaring Ralph.

(for those who missed it, scroll up for my picture)

Get well, Ralph.

Kev: mind adding me to board list?

ralphm, You know an eyepatch is meant to be black, right?

Best they had

dwd: I'm missing a parrot.

Also pretty sure most pirates had a white one first

Ok, I'm off. Car / parking lot / cold.

Mail me so I've got the mail address in my inbox.

Kev: my Gmail please

See my application

Oh, also congrats on council and dwd as chair

(of council)

ralphm, Ta!

I think a single talk slot won't suffice any more at the upcoming Summit. Too many problems in https://op-co.de/tmp/whats-wrong-with-xmpp-2017.pdf

oh, you updated it

maybe we need to have the summit last one more day and give one full day to Ge0rG

mathieui: my immediate reaction was to veto that proposal, but then I realized it is for the Board to decide, not for the Council.

I think (keeping in mind that I've only been to a handful of summits) that summits work best if you pick a specific issue and focus on that

SamWhited: I could pick out Message Routing and fill a 30~60min slot with that probably.

That does seem like a big one

SamWhited: it's also the most important one, IMO

SamWhited: after that is tackled, a new client session type or some kind of MAM-sub can be applied to solve the remaining minor practical problems ;)

I also actually wanted to create that message routing rules table for a long time now.