XSF Discussion - 2021-05-20

alright, took me about 300 lines to make the domainpart parser. Not too bad lol

for just the parser, that sounds like to much LOC

for something that prepares, validates and enforces the domainpart, it is way to less (assuming no external libraries)

definitely leaned on many libraries for it, and it's just the parser (which by definition validates btw). I didn't do enforcement, because I actually don't know what the spec means by that. I think I just do that when I'm going to do comparison, or if I'm going to store or forwards the information. ✏

depends on your use case, servers to a lot of jid comparision and "is child of" checking. so it appears sensible to store the enforced variant of the JID representation

What /does/ enforcement mean?

IIRC lowercasing the localpart for example

So convert everything to the normalized form and reject invalid values?

I'd assume rejecting happens in the validation step

We occationally have users who REALLY want their UserName case to be preserved. :)

I'm sure you do

hmm it appears in PRECIS, validatation and enforcement are in the same step "enforcement"

qrpnxz, I am sure you checked the RFC's errata for additional information regarding domainpart validation, right? :)

i have not finished reading the spec even :) Thanks for heads up tho

enforcement seems to be about taking an almost valid jid (say from user input) and doing some transformations on it so that it's actually conformant ✏

yeah i'm pretty sure about this see RFC 5895

qrpnxz, what about ipv6 addresses in the domainpart? maybe there are code points that IDNA disallows that are valid in an IPv6 address?

IDNA is not relevant to IP addresses, only if you detect that it must be a domain name do you start worrying about that

flow,

right, but you should not allow arbitrary strings if it is not a DNS name

and, how do you detect that is must be a DNS name?

if it starts with a number it has to be ipv4, if it starts with a '[' it better be ipv6, otherwise i start trying to parse a domain name

DNS labels can start with digits, so that's not sound

LDH cannot start with digits

U-labels maaaaybe. And maybe when i find time to care i change the code to try domain name even when it starts with a digit. That would not be hard.

yeah no it can't lol

it gets caught in the bidi rule

> LDH cannot start with digits source?

sure one sec

https://datatracker.ietf.org/doc/html/rfc1034#section-3.5

they also cannot start or end with hyphen

https://datatracker.ietf.org/doc/html/rfc5890 didn't say so, only the hypen bit

I thought that was relaxed in a later RFC.

yeah 5890 defers to 1034 on a lot of details

(Starting with a number, not the hyphen)

But I don’t have this swapped in.

1123 maybe? I don’t have time to read it to check.

Blah. I got sucked in

2.1 Host Names and Numbers The syntax of a legal Internet host name was specified in RFC-952 [DNS:4]. One aspect of host name syntax is hereby changed: the restriction on the first character is relaxed to allow either a letter or a digit. Host software MUST support this more liberal syntax.

hmm, nice find. Problem is that it breaks the bidi rule

Kev, which are you quoting from

1123

wait, but this is for host names, is that really the same as a domain name?

Down this path madness lies. Don’t say you haven’t been warned :D

also, a host here can be an ip or a domain name, so ofc a host name should be able to start with a number, but this is not necessarily the case for a domain name

yeah i'm certain this is what that's for

it's confusing, isn't it? rfc5890 even explicitly says that LDH label rules are modified by rfc1123

> yeah i'm certain this is what that's for Honestly, if you’re still at the point of being certain about anything you’ve not spent long enough looking at name stuffs :D

I tend to lean towards rfc1123 § 2.1 talking about "DNS labels" when "host name syntax"

the problem is, besides what Kev mentioned, terminology

Kev, correction: I'm certain this is how i'm going to interpret it xD.

"domain name", "host name", are not well defined terms

"DNS name", "DNS label" are

Actually, domain name and host name both *are* well defined, I think. The former in 1034 and the latter in 1123. It’s just that … yeah.

even the JID standard just tells you about RFC5890 in an informational note. The actual requirement is that it be a FQDN, which the standard also admits as not well defined lol

It’s defined what format they have, but not what they *are* :) ✏

Kev, not sure if this is the case for "domain name", rfc1034 even admits "The terms "domain" or "domain name" are used in many contexts beyond the DNS described here."

Fair.

where "DNS name" is unambiguous

But frequently unhelpful!

of course, you could argue that it is well defined to be not well defined

xd

qrpnxz, would you mind elaborating on the bidi rule thingy you mentioned?

yeah so for full unicode labels there's all kinds of rules for what code points are allowed, and one of them has to do with the directionality of the text (left-to-right, right-to-left). A label is only valid if it passes this test. Link: https://datatracker.ietf.org/doc/html/rfc5893#section-2

and having a digit at first position causes issues with the bidi rule because?

(note that I may have misunderstood you)

well it violates condition one, for starters

right

so you can't have a U-label starting with a digit

but you can have a A-label

so 1soccer.us would be fine as DNS name and as XMPP domainpart

but 1fußball.de would not be fine as DNS name (but I believe it would be fine as XMPP domainpart)

then again, a XMPP domainpart that does not map well to a DNS name is mostly useless ✏

A-labels start with xn-- remember :)

no

yes

no

this is just fact sorry

quote "This means, by definition, that every A-label will begin with the IDNA ACE prefix, "xn--""

got me, was confusing A with LDH labels

and LDH cannot start with number for sure

and that's because?

https://datatracker.ietf.org/doc/html/rfc1034#section-3.5

because that's how it's defined

but everything is a LDH label, so that relaxation from rfc1123 would apply there too ✏

i'm really certain that that quote from 1123 is for host names, which it says were previously defined in rfc 952 (totally different document)

but why would rfc5890 then refer to rfc1123?

also rfc1123 talks about DNS' "trailing dot"

but yes, I can definetly see your point too ✏

https://datatracker.ietf.org/doc/html/rfc3696#section-2 add any light here? :)

moar rfcs!!!!

i refered to 1123 because you brought it up lol

but 5890 refers to 1034 on what a valid LDH label is

Kev, I think it does, and I am pretty sure I've read all that ages ago

the second paragaph only mentions the hypen rule, nothing about disallowing a digit in the first position

my god everything is mislinked but i found something one sec

"a segment of a host domain name is now allowed to begin with a digit and could legally be entirely numeric (see Section 6.1.2.4). However, a valid host name can never have the dotted-decimal form #.#.#.#, since at least the highest-level component label will be alphabetic" this is from 1123 section 2.1, twas hidden later on in the text. (Section 6.1.2.4 is totally completely unrelated btw) So the answer here seems to be then to only apply the bidi rule if there are any RTL characters. (I'll have to check if there was a correction of the rule.) ✏

this i think is better evidence that it also applies to domain names in general

> Honestly, if you’re still at the point of being certain about anything you’ve not spent long enough looking at name stuffs :D I’ll just leave this here :)

real world proof of number domain name https://37signals.com/

will get back to you guys on the bidi thing

Oh, I could have come up with exxamples if asked :)

bruh lmao

123-reg is the most obvious one I think.

you bastard, anyway, i guess for bidi it's like i said, only applies if there's an RTL somewhere in the whole domain name

I thought I started off by saying it was allowed ;)

i mean idk why i didn't even try, clearly there's 4chan and 2ch

hahaha

qrpnxz, that bidi thing only applies to U-labels, no?

so you could have a plain old LDH label with a digit at first position, but not a U-label

domain names with RTL characters, it says

a U-label could have it if no other part of the domain had an RTL char

isn't that all something scoped within the context of IDNA / U-labels?

necessarily the label with the RTL char would be a U-label if that's what you are talking about, but the bidi std says that the rule applies to all labels in a domain with any RTL chars

that's now what I remember, expect or have read, but then again, it's big area

sorry, to be more accurate, that contains a char with bidi class of R, AL, or AN

IIRC the bidi rules only applies to U-labels

and I'd argue that todays DNS is not all U-labels (but could be wrong)

it says "The following rule, consisting of six conditions, applies to labels in Bidi domain names." Which are domains with at least one RTL label. Which are labels that contains at least one char with bidi class R, AL, or AN

it being?

anyhow, isn't foobar in foobar.com an example of a non RTL label?

(hmm what is bidi class AL…)

AL is arabic letters, AN is arabic numbers

R is most letters is non-arabic RTL scripts

foobar.com is not a bidi domain name

that's what I am saying

neither foobar nor com are RTL labels

so the bidi rule does not apply here

correct

which means I can have 9foobar.com

correct

ahh, now I understand what you said: not only does the bidi rule not apply if it isn't a U-label, but I can have a U-label without codepoints that would require the bidi rule

yes :)

9fóobar.com is also fair game

like i said, it applies if it's a bidi domain, which means it has at least one RTL label, which has at least one type R AL or AN char

which 9fóobar.com does not have

<qrpnxz>  it gets caught in the bidi rule made me believe that you thought that the bidi rule would disallow a digit in the first place in every U-label ✏

turns out no

https://ru.wikipedia.org/wiki/XMPP_Standards_Foundation

Amazing, such a long russian wikipedia article about the xsf.

😲️

Jabber is big in Russia

That's why we have millions of $ of funding from the Kremlin to add security issues in the protocol

🤦‍♂️️

Unlucky they sent it to the wrong BTC address.

Perhaps we should change policy to accept rubels also.

It's a little out of date. It says Кевин Смит is still on Council.

It is amazing that they care about the names.

Who the hell finds it interesting to have a membership list on wikipedia?

dwd: And I always will be in spirit :)

Kremlin didn't come to me yet =) but FSB (ФСБ/FSS) did.

The German Federal Cartel Office is doing an investigation of chat services, and they would like to inquire the operators of "Jabber", in German, via snail mail. Can Board empower somebody to respond on behalf of the XSF? I'd volunteer, unless Daniel wants to step in.

https://www.bundeskartellamt.de/SharedDocs/Meldung/DE/Pressemitteilungen/2020/12_11_2020_SU_Messenger_Dienste.html

Is board the right entity given that board does not represent a chat service?

That said, the XSF should try to be present in this discussion

but clarify that there is a difference between "chat service" and "chat protocol". Maybe on the example of email

"distributed open-standard chat procol" that is

flow: I'd take care of that

I'd expect nothing less :)

so some XSF representative *and* another represenstative from a XMPP-based "chat service" would be ideal

hence why not appoint Ge0rG, and send daniel?

(given that he wants to)

what do you mean by that?

why not write it and share it here?

before "submission"

eevvoor, I don't think anyone would mind that. but first board has to decide/delegate

flow: a few people in the xmpp verse have been requested to participate on their own behalf

Meaning there is probably already some diversity in it

Daniel, ahh, ok, that sounds good then ✏

I know of inquiries to Jitsi and BBB

weird

I wouldn't make too much of a big deal out of it. Just have Ge0rG do it if he wants to

argh i accidentally cleared this chat as i was reading the juicy part

qrpnxz, https://logs.xmpp.org/xsf/2021-05-20

wow life saver

We've a Board meeting this evening, so we can rubber-stamp Georg if that works?

> and another represenstative from a XMPP-based "chat service" would be ideal how about a random self-hoster xDDDDD

dwd: that would be great, but please be gentle with the rubber-stamp

yeah, a silicone stamp is more body-friendly

do you know why the "German Federal Cartel Office" is doing a chat service investigation?

does the ink contain lead?

Investigation of what?

Branding? Tatooing?

qrpnxz: I've linked their press release above, maybe some google translate might help

Ge0rG, Is tha "Cartel Office" the equivalent of the Monopolies and Mergers authority?

Ge0rG, actually there's an en version https://www.bundeskartellamt.de/SharedDocs/Meldung/EN/Pressemitteilungen/2020/12_11_2020_SU_Messenger_Services.html ✏

qrpnxz: it's probably rooted in the monopolies of Facebook and such, and they want to see how to ensure interop. This has been going on for a year or two now

qrpnxz: cool

looks like they are investigating user data protections

> Furthermore, in most cases it is not possible for users to communicate with each other by using different messenger services. GO XMPP

dwd: I can't possibly answer that without knowing what the latter is about, and it looks like the M&M was replaced at least twice since.

the name *does* sound like it could match the kartellamt tho

btw, wikipedia: " Federal Cartel Office (German: Bundeskartellamt) is Germany's national competition regulator." So "Monopolies Authority" sounds about right.

Mexico's Federal Cartel Office is quite different

i think it's https://www.cofece.mx

covfefe

It is about that time again

That comes quick

Yeah weekly meetings do that

0. Welcome

Hi!

Who do we have today?

o/

Any point for our agenda?

Here

Yes, the Ge0rG / German Cartel thing mentioned earlier today.

dwd wanted to rubber-stamp me.

Missed that, but ok.

1. Minute taker

I took meeting minutes last 2 weeks

Yes. I think that's dwd now

OK

2. "rubber-stamping Ge0rG"

Ge0rG, Want to explain?

Sure!

The Federal Cartel Office of Germany (Bundeskartellamt) is inquiring different IM providers about interoperability and similar topics.

They have also approached the person behind freie-messenger.de about a contact for "Jabber" in the context of chat and video calls.

I suppose that they mean XMPP and not the Cisco product, and I think somebody from the XSF should step in to participate in their questionnaire.

I would step up to do so on behalf of the XSF

They require an address in Germany and some hopefully small time effort.

So, I think this is a fine idea, and if Ge0rG is happy to do it that's great. I think Board should be sanctioning "speaking on behalf of the XSF", but in this case I feel it's a formality.

So I would like to get rubber-stamped by the XSF Board as the Official Ambassador Of XMPP For The German Government.

heh

Sounds good to me

I think we've previously used the word liason, but in general I'd be happy to have Ge0rG be the in-between.

I think it should be "The XMPP Ambassador to the Federal German Republic", technically, but whatever.

(For the avoidance of doubt, I am joking).

Or rather "The XSF Ambassador...", as I've read people be sensitive about this even yesterday.

dwd: I was just going to ask for a diplomatic passport.

I take dwd's thinking as a motion.

+1

+1

h+1

arc?

dwd: technically it would be the "Federal Republic of Germany".

+1

Thank you very much, I'm honored to represent the XSF in #Neuland.

Ok so who's making him the officially xmpp loferl socken?

Motion carries. Congrats!

For the minutes, you can link the slowly loading page at https://www.bundeskartellamt.de/SharedDocs/Meldung/EN/Pressemitteilungen/2020/12_11_2020_SU_Messenger_Services.html

arc: I believe you just volunteered

Yay!

3. Open Collective

From what I've seen Peter was dedicating an hour on this earlier this month. I haven't seen an update, yet.

Sam pinged earlier today.

4. AOB

?

Not this week

arc: but you have one for next week?

I'll wait in anticipation.

Maybe. We'll have to see

5. Data of Next

+1W

6. Close

Thanks all!

Ge0rG: medium shoe and calf size? 😏

Is that the new ASL?

For the Loferl Socken!

Thanks all

arc: sure

As long as I'm not supposed to wear a Dirndl.

Just a hoodie.

_just_ a hoodie?

No it would be great to get you in a photo wearing xmpp socks and the leather pants

arc, Now you remind me I need to work on that CoC...

But seriously just yesterday I was talking to a friend about loferl socken to be worn under rugby socks because they look super warm, and as long as they are less than 5mm thick they would not violate any rules.

I have known a few people that have worn foam shin pads under their socks. But yeah anyways I was downloading knit patterns to make them

Minutes sent. ✏

Isn't the warmth simply because they're 100% wool and roughly knit? ✏

And they are literally anything around the calves. Rugby socks are typically knee-high but also typically thin so they are not all that warm

I have played in the snow. So I can testify that they are not that warm. As can all of the bonfires with players huddled around them

I seem to recall that was just considered character building.

So basically you need some kind of Merino over-the-calf socks

In my experience is not very. When we were playing in Manchester England it was pissing rain at around 2c out, and windy, and I had a miserable weekend

Or, you know, just not play rugby when it's cold.

ralphm: that's what Loferl Socken is

arc: note that Merino

That's a brand, isn't it?

Also, Manchester, 2°C and heavy rain - so July or August? The rest of the year it's windy, too...

arc: let me know when you have my XMPP branded Loferl ready, and I will arrange for a photo in leather pants.

It is a breed of sheep with certain wool quality, and you now often see it used as a type-of-wool. Often thin but warm.

Might make it easier to search

dwd: if people wouldn't play rugby when it's cold, how did it ever take off in England?

ralphm, I spent much of my childhood asking the same question.

dwd, those are fast minutes!

haha

ralphm, Usually while being forced into playing Rugby.

<Arc> the XSF is too European. <also Arc> I demand our liason wears loferl socken.

:D

*boom*

Is xmpp.org especially slow tonight?

seems fine to me

https://xmpp.org/software/libraries.html has "project name: txmpp, languages: linux" ? it's neither a library nor is it written in a programming language named linux...

moparisthebest, we went through this already, I think someone needs to put the work in adding a new page for "tools" or whatever

(which txmpp is)

it looks like a sendxmpp that relies on libstrophe from the link

looks like it's already on this "tools" page https://wiki.xmpp.org/web/User:MDosch/Sendxmpp_incarnations

this one seems more like the library: https://github.com/rpavlik/txmpp

interesting, seems abandoned though

The language would be "Unix shell"

the one on libraries.html is actually written in C

"Linux" makes no sense regardless

what in the world https://github.com/rpavlik/txmpp/blob/master/src/Equifax_Secure_Global_eBusiness_CA-1.h

Applogies for the random message. I was told that I should ask this here to get a precise answer. I am looking for fundamental differences between XMPP and MATRIX. I'm sure quite a bit has been written already. If you have an exhaustive link to point me, that would be great.

If you just have some bullets for me...I'll take it 🙂

there are a lot of major differences, I can list a couple: 1. XMPP is an established standard with RFCs that has passed the test of time, Matrix is a... VC funded project? 2. XMPP is a message passing protocol, Matrix is some... eventually consistent graph thing? This is why matrix servers take insane amounts of resources

eventually consistent globally replicated JSON graph database

> eventually consistent globally replicated JSON graph database Don't know if it's just me, sounds complicated.

roro, also XMPP is XML over TLS, Matrix is JSON over HTTPS, that's more on the level of implementation detail that doesn't matter though

dwd, ralphm arc I also offered Ge0rG help if I can do something. And I also followed the process so far. And some other members here too.

> jubalh escribió: > this one seems more like the library: https://github.com/rpavlik/txmpp We allocated this to Libaries/Tools on the website I thinl

https://xmpp.org/software/libraries.html