XSF Discussion - 2019-09-12

Good morning

good morning

are you *the* daniel?

No, he's Derda Niel

would the XSF want to host blog posts showcasing e.g. how to configure a specific XMPP setup on xmpp.org?

from non-XSF members

jonas’, maybe not 'blog post' but actually good tutorials i think are welcome

re the discussion on the developers foundation at the last summit

maybe we can just retweet them from @xmpp

when ever i finally have a few days down time i'll actually submit such a tutorial myself

here is how Conversations is going to deal with the multiple translation issue from now on: https://github.com/siacs/Conversations/commit/9bf5fb98acce52d49326f2f613a99f5b957f95c7

Daniel: got a screenshot of the language tag?

Daniel, is there such a thing as "no language tag"? xml:lang is usually (often? always?) set on the stream

https://twitter.com/iNPUTmice/status/1172067975354114048

pep., no tag means stream

and then Conversations will show that

I type in Dutch and English depending on which contact I'm talking to. I am Dutch, but have my phone set to English. How do you plan to handle this?

I also do type Japanese on my xml:lang="en" stream fwiw

ralphm, it's just displaying the tag if there are multiple bodies

so you will never see that anyway

I mean, how do I tell my client what I am doing?

Most clients don't or badly support multiple languages anyway

right. but Conversations will not send multiple. if you have a client that sends multiple it is the responsibilty of that client to set something sane

i mean if you type a dutch message with Conversations it will only have one body and thus not show any tag on the receiving side

But even if you have a single body, would you then just not have a language tag, and just have it inherit from the stream?

ralphm: change the PC / mobile phone language setting before talking, reconnect to update the stream language

:D

Ge0rG: of course

08:50:07 Daniel> pep., no tag means stream

Daniel, no tag means the tag on the stanza

well no tag means parent tag

and so on

Well xml:lang is inherited as defined by the XML spec, so that part was clear to me.

no single body will not show a tag

that would be extremly annoying

Daniel, overall, that sounds like a very reasonable solution

It's not inherited in most xml parser implementations...

Ge0rG: yep, it is not easy, and then there's the part where you take a stanza from a stream and send it onwards on another stream. You'd have to take the inherited xml:lang along (just as with namespaces/prefixes).

(i'll make a note for looking into the inheritence)

I'm still not sure if having multiple language bodies is useful at all.

Daniel, inheritance applies to other things besides xml:lang too. If you are going to fix it for xml:lang, you may want to make sure that it is also done (or at least possible) for those other things. At least that is what I did in Smack

ralphm, what about e.g. pubsub notifications to a wide audience?

flow, what are the other things?

I only know of xml:lang

wasn't xml:space a thing?

if that isn’t handled by the XML parser you’re screwed anyways (and I don’t think it’s inherited?)

> I'm still not sure if having multiple language bodies is useful at all. As I argue in the commit message just showing the default without any warning doesn't help because you never know what legacy does out there

So whether or not sending this in practice is a good idea is only a secondary question. Your client will have to deal with it either way

jonas’, namespaces and prefixes

flow, right, I assume that everyone uses an XML Namespaces compliant parser

if you don’t, you’re in a world of pain anyways ✏

jonas’, I would't agree with your assumption and conclusion, but yes, makes things easier

flow: for pubsub, yes, that might be a good idea. But I thought we were mostly discussing regular messages with the standard body element.

For regular conversations, denothing the language is impractical, at most.

And at the same time language is hard. Especially if you look at far-east languages. Chinese, Japanese, and Korean, can all be expressed in Unicode. However, due to the so-called han-unification, it means that they all use the same code points with (wildly) different glyphs. So there, denoting which language it is, is crucial for proper display. I am unsure how this actually achieved in platforms.

probably from context? ✏

right-click, select encoding. Like in the browser in the good old days of koi-8r

Ge0rG: it has nothing to do with encoding

You may want to read about Han Unification.

ralphm: so it would be: right-click, select *de*coding

that's not related to xml:lang anymore is it. Sure language is hard

ralphm: is it something to be covered by xml:lang regional selectors, or by unicode variant selectors?

pep.: well, it is related in that you arguably could use xml:lang for selecting which variant glyphs to use.

ralphm, you mean automatically?

Ge0rG: and yes, variant selectors might help, too

pep.: maybe there are heuristics, but I don't know

is there an xml:lang that will render all gender-neutral emojis as female?

haha

Ge0rG, no, there has to be a male modifier alongside!

(that's what I see most often anyway..)

xml:lang="en_US@♀\u1F3FF"

Those are ZWJs, not variation selectors, I think?

ralphm: ZWJs are used to append variation selectors, aren't they?

The number of "characters" in a single Emoji is a highly debatable thing.

https://hsivonen.fi/string-length/

Ge0rG: I am a bit out of my depth here, but I think that using presentation selectors with emoji (with ZWJ) is different from what unicode calls variation selectors.

I think variation selectors (regarding emoji) is for making a distinction between a monochrome glyph and a colorful one.

ralphm: right; I've totally lost track of the different selectors and how to glue them to the actual Emoji

I think this page explains it nicely https://blog.emojipedia.org/emoji-zwj-sequences-three-letters-many-possibilities/

It also mentions VS-16, and I was correct, it seems.

Apparently, you just append the fitzpatrick modifier, but use a ZWJ for the gender.

(under the heading Twitter, with a footnoot 3)

Now I wonder what happens if you strip out VS-16 from the woman-heart-woman or the facepalm-man emoji.

I am still hoping that some day, having WAFFLE + ZWJ + REGIONAL INDICATOR N + NATIONAL INDICATOR L will yield a Stroopwafel.

You probably get a monochrome heart

maybe

Inside the colored Emoji?

this makes it explicit

I don't know man

Can I use a 💔 in that sequence?

So many questions.

https://twitter.com/ralphm/status/1062658994341339137

I installed a font that supports Unicode 12, now, so I see a waffle at least.

(I didn't at the time)

ralphm: does it have the epileptic parrot Emoji?

(The Epileptic Parrots would make a great metal band name)

It is called party parrot

https://cultofthepartyparrot.com/

crazy

There used to be a parrot.exe bundled with certain SoundBlaster cards. It had a cult as well. https://m.youtube.com/watch?v=DCMJ0Ufiilg

:D

That is awesome!

ralphm: there is a bunch of covers and even some live performance videos!

am i reading 313 correctly that it doesn’t warn you not to make the dino mistake?

or am I missing something?

(matching up query-id and service jid)

I don't recall any text along those lines, no

I agree that evidently it needs to be documented

But I would have assumed it was obvious that results would come from the JID you queried

and not from a different JID

well…

To be clear, I'm 200% in favour of documenting anything that will help people write more secure software by default

Just explaining why it probably wasn't added in the first place

sure. to be clear i found it 100% obvious to match the jid and the query-id

I found it obvious as well, but my underlying library was playing games with me.

I think it's obvious to everyone if they read it properly, but it's easy for anyone to forget things, so it's good to make important things explicit

Smack is calling the processMessage() callback on each message, including on messages that contain a MAM payload, and it is actually the straightforward way to just extract and use it there.

But processMessage() will also get called on all malicious MAM injections.

So even when you are aware of the pitfalls, it's not always obvious

Ge0rG, which call site of (assuming the MessageListener's) processMessage() are you referring to, from MucManager?

flow: for historical reasons, I'm going through connection.addSyncStanzaListener(..., new StanzaTypeFilter(Message.class))

flow: there was a time when smack was silently removing your listeners on network events, like a disconnect

so I had to work around all those

Ge0rG, so you tell smack to invoke a listener for every message and then appear to be complaining about it doing so?

flow: looks like that, yes.

flow: what's the official way? A ChatManager?

What do I do to track message delivery errors? A ChatMessageDeliveryErrorManager?

0. Welcome + Agenda

Who do we have today?

Seve sent regrets.

I still have network, I can follow for now 👍

You might need to incant their names.

Hey

It worked

I didn't even incant, yet?

test

ralphm, Negative latency.

dwd: figures

Guus?

I assumed you did it out loud

Guus cannot attend

Trello as agenda. Any additional items?

None here

Seve: I missed that, ok.

re-re-test

nyco: works?

There are awaiting PRs in the repo. I was wondering if that should go through board, but I guess I'll poke jonas’ first

shaky connection

I'm here, but maybe not

pep.: PRs for what?

1. Minute taker

Who?

i can do that

Thanks Daniel

2. POSS

nyco put this on the agenda

oh yes

actually maybe rather SCAM

ralphm, DOAP for example. but ignore it for now, I'll try to get that in the agenda for next week

nyco_ that'd be my first guess, yes

ok

3. ralphm's items

I have been occupied with other stuff, sorry.

Will try for next week.,

4. GSOC

I saw this was discussed last week?

yep

mentors & flow have no time to write stuff...

With an apparent action item for commteam to write a blog post

commTeam may aggregate students blog posts...

Can you take it up with commteam?

sir, yes sir

Much appreciated sir.

And shine your shoes next time.

:)

sir, sorry sir

5. Adopt a character

Flow suggested this one

Besides the confusing terminology – I assume they mean codepoint – why should we do this?

because it's funny and only 100€

we could attract some sympathy

apart from that it is useless, that is why it is absolutely necessary

Surely funny.

I also should ask: should the XSF put money (from its sponsors) in this?

good point, but that's only 100€

Silver sponsors pay 1k€

I don't really know how much this is known though, but I share the same opinion as Guus and Nyco. We can poll the membership as well, or gather more information on the popuarity of it, etc. Ideally we could promote ourselves a bit as well.

we have some money in the bank, right?

It's even less, because USD, and tex deducable. And it's for a non-profit org.

ok, less than 100€

I think the popularity of this thing has nothing to do with what we can do with it

I'd also hoped that one of your sponsors would help us for the silver sponsoring, but even 100$ would be ok and increase the visiblity of the XSF a bit.

I have mixed feeling about it, I defend the idea now, but I'm not so sure, just feelings

My general feeling is that it could be justified to spend 100$ for this. But I have no detail knowledge about our bank account

We could have the badge on our website which is cool

s/badge/character/

:)

No, they generate a badge, Nyco

We could also ask members to help with the funding, if it's a problem

I also should point out that Bronze level is not unique, that's only for Silver and Gold.

the badges for compliance are coming which will improve our visuals

ralphm, IIRC only gold is unique

but silver limited to 5 or so

oh, ok

while bronze is unlimited

https://xmpp.org/community/sponsorship.html no bronze

not this bronze

oh waut

ralphm, re the terminology, "adopt a character" is how unicode.org calls it

flow: do you know by any chance other communities that did this?

but anyhow, appeared like a cheap way to get some visibility and it's for a non-profit that we benefit from, soo

Seve, just look at the current sponsors

https://www.unicode.org/consortium/adopted-characters.html

Yeah, we are not Adobe :D

I know of one company that went crazy with this, by the way: https://www.elastic.co/blog/elastic-love-for-the-adopt-a-character-program-and-the-unicode-consortium

instead of sponsors money, indeed we can even make a small crowdfunding campaign, which brings visibility as well, allows us to measure somehow our attractiveness

flow: my 😜 wasn't obvious?

not sure how visible you really are among 600 other bronze sponsers. that feels like a go big or go home situation (where you really want gold). and gold feels too expensive for us

ralphm, Is that the emoji we're adopting?

> not sure how visible you really are among 600 other bronze sponsers. that feels like a go big or go home situation (where you really want gold). and gold feels too expensive for us That's my "issue"

I agree with Daniel

dwd: this is not my suggestion, no

it is not about visibility among sponsors, it is about us generating original content and gettings relayed and engage the communities

But if you'd want to do it in a big way, you'd adopt < and >

At Gold level

hehe

the chat bubble at bronze level will do

My view, BTW, is that adopting is worth a punt at the $100 mark, and we can make some noise about it ourselves. It's just an interesting talking point at that level, but a talking point is always useful.

My biggest fear is that one of your competitors shows off by adopting more and expensive characters

we communicate on our channels, our competitors don't come on our channels

I like the idea and would love if somebody among us was motivates enough for this, to write a blog post about how much we like making a donation and adopting a character. For me is really a +1, just that I don't find myself inspired enough about the whole unicode world to promote it properly

also, let's focus on ourselves, our future, how we serve our people

if you want something to tweet pay for drinks and/or dinner at the next developer sprint and we get you pictures of happy developers developing xmpp

looking at competition is a waste of time, makes you feel bad/good

that can be done anyway

dudes drinking beer is not funny nor attractive btw

> My biggest fear is that one of your competitors shows off by adopting more and expensive characters Surely they can afford it, so yeah, but we would be first as well

> to write a blog post about how much we like making a donation and adopting a character. Tbh I see this easily criticized.. "XMPP don't have anything better to do? They could fix their protocol instead"

(Not by Matrix obviously)

there, you named it! :)

the same could be said to any char sponsor

nyco_, yes

what do people actually say on Elastic for example?

quite sure we don't care about _potential_ negativity

pep., You realise people say that stuff *anyway*, right?

Exactly

indeed

dwd, yeah

I see enough support for asking the community what character they'd like to see adopted

let's make short blog post, a call for funding, we offer the bubble bronze, that's 100€ we see how the community responds to this we move on, or not if we do, a second blog post to announce the bubble is ours in a funny way

🐑

Kev: that's you suggestion?

how does it work to collect donations from the community?

ralphm: Well, ask a Welshman :)

Dude we have Texel

I'm still in favor of the light bulb

I'm not

Guys, I'd humbly suggest you just pick something (chat bubble sounds good) and do it. No real need to ask the community.

is there a bike shedding emoji?

not lighbulb, it's Jabber, it dates from the XXth century

Correct

The rules are here: https://home.unicode.org/adopt-a-character/sponsorship/

The irony: https://matthewwild.co.uk/uploads/screenshot-20190912-1568296922-7917.png

symbolises presence, which has become overtime a secundary feature

I like presents

dwd: thank you

MattJ: https://twitter.com/ralphm/status/1172092470311739392 should help you

MattJ, Could we sponsor U+FFFD then?

hah

?

+1 on selecting the chat bubble, but noting it is also sponsored at Gold level by Discourse

+1 chat bubble

dwd: why not ALM?

Discourse, we don't care, these are friends, not competitors even competitors, we don't care, see above

Are there any other chat bubble characters? I'm guessing there is no "real time" emoji

U+1F4AC 💬 U+1F5E8 🗨 U+1F5E9 🗩 U+1F5EA 🗪 U+1F5EB 🗫

U+003C?

we could list candidates, but the chat bubble is pretty obvious and expecetd

dwd: :D

dwd

I suggested < and > above

given the popularity of XML...

I suggest (as Dave) that if Board want to do this, just come up with something and do it :)

Make XML great again!

I like 🗪 U+1F5EA

Zash: thanks for the swag idea there

I still think we would have to make it a bikeshed emoji, but there are no bikeshed emojis, and the closest term would be yak shaving, but there are no yak emoji, only water buffalos.

Ge0rG: Can we propose a new modifier for the water buffalo?

ralphm: it's nice, except that it's not yet widely supported

But FWIW, I think ralphm's suggestion of U+1F5EA is cute.

Ge0rG: well, we are progressive

And it's nice to have something that if we include it in JIDs will break the network ;p

> Copy 🚩 This Unicode character has no emoji version, meaning this is intended to display only as a black and white glyph on most platforms.

ok, it's 16:09, the meeting has been sabotaged by the floor! :)

nyco_: ehem, somebody asked for input by the community

But I agree we are done for today.

oh...

I don't know who

Board: let's all think about our preference and hammer it next week.

For those without a decent font, like me: Unicode Character 'TWO SPEECH BUBBLES' (U+1F5EA) https://www.fileformat.info/info/unicode/char/1F5EA/index.htm

dwd: see my tweet for help

ralphm: your help is not universally applicable ✏

6. AOB

nope

None here

Ge0rG: you could at least look at it in more detail, and see that the second link actually is for the most part.

7. Date of Next

+1W

+1

8. Close

Thanks Daniel for the minutes.

And thanks everyone else, too.

Thanks

Thank you for the meeting, got a bit off topic though :D

thanks, merci !

what's 'POSS' by the way?

Daniel, Paris Open Source Summit

Paris Open-Source Summit

i see. thanks

isn’t that scam then?

shouldn’t we at some point merge scam and comm?

and make it 'outreach' or something?

Yeah I'm of this opinion as well

Daniel, There's certainly a lot of crossover.

shouldn't that be decided by the respective teams?

Well, I generally just do SCAM, and I prefer that

There is no reason why work teams can't work together

I'd also prefer it to be able to just set these teams a budget, and let them discuss unicode character options for hours

Indeed

That seems rather sensible.

I think it's somewhat wasted time for board

SCAM already has a budget

It feels like a lot of trivial stuff trickles up to Board these days

we have a budget?

TIL

Ooh, XSF trivia Kev didn't know.

Yes, we set one for SCAM

oh we do

cool.

FYI last night we got paid for GSoC, so we now have an extra $4800.00 in the bank account. Don't spend it all on Unicode characters! ;-)

It is $1000/yr

peter: hehe

ralphm: for gold?

now we have to spend it all so we can get it raised for next year (that's how budgets work, right?)

indeed

"See, we need moar"

"You have money left? Better reduce next years budget by that then"

that should be enough for Gold on <🗪>

i.e. all three of them

no wait, just for silver :(

See, *do* need more.

I would actually be very happy to see more money spent on outreach in general.

Zash: for SCAM. https://mail.jabber.org/pipermail/members/2017-September/008648.html

pep. and I actually wanted to spend money on scam a couple of times. but we didn’t know we had a budget

Ah

dwd: I offer to loudly proclaim "XMPP" in the shopping mall in my home city for just 250$/hr

I'm not sure I would do that even at this price

Daniel: well, every time people have asked about money (like for stickers) I said: yes, send the bill to peter

pep., words true of real XMPP developer :)

:D

pep.: I was trying to think of the lowest-effort thing that qualifies as "outreach for money" that doesn't immediately qualify as a SCAM

ralphm, sure. i just wasn’t really aware that those came out of the scam budget

Ge0rG, Unicode characters?

Let's start a "sponsor an xmlns" campain! /notentirelyserious

dwd: this results in being placed on some well-hidden web-page. We can have that already for free.

Daniel: well, we haven't really been strict and make a financial overview

most of the money is spent on the Summit / FOSDEM and that $1000 is a nice target

14:03:38 dwd> MattJ, Could we sponsor U+FFFD then?

stpeter: if you are kinda here, have you seen my mail over at xmpp@ietf.org regarding Unicode versions?

I literally laughed out loudly, in a fiendish laugh. Thanks for that one.

ralphm, I don't see it. When?

dwd: https://mailarchive.ietf.org/arch/msg/xmpp/a-WhzOTyOq168GujQHgzQ1-DURI

I've found it now.

dwd: wow, if you missed it then I guess nobody's read it

ralphm, I have a eyeball-meltingly large inbox, mind.

Select All, Mark as Read.

Select all, Archive.

Read Mail Really Fast.

Inbox Zero Archivement

> =inbox [Msgs:163039 New:99724 Flag:1477 Post:11 Inc:20 7281M]

ralphm: I did see your note among the other ~10k unread messages. ;-) I'll need focused time to spool up all that i18n context again, hopefully tomorrow.

bbiab

peter: 👍

flow: so, what's the official way to process messages? A ChatManager?

What's the response to a MAM request which matches no messages?

`<iq type='result'><{urn:xmpp:mam:2}fin complete='true'/></iq>` ?

{rsm}set/count=0 maybe

Don't think it's explicitly stated anywhere

https://github.com/xsf/xmpp.org/pull/409 should this be to board? Is seems there is no place on the website where CC or Creative Commons is mentioned apart in specific places like IPR changelog, or a blog post I can't find the url for

*sent to board

Also as mentioned earlier in the board meeting, https://github.com/xsf/xmpp.org/pull/594 DOAP. Maybe Link Mauve can chime in next board meeting if we put it on the agenda

(Is all that handled by board, or anybody else?)

pep., Licensing is Board, DOAP... Might be? Probably is, but they may bump it straight to whoever is looking after the website.

Who is looking after the website?

(Editors surely have commit rights, does that make them "looking after the website"?)

I don’t think that editors have commit rights per-se?

I suspect it's Board

pep. add PR 409 to the board Trello board

thx

I don't have a trello account, can you put it there in the meantime(?)

you put it there

oops

my mistake

will do

> pep. add PR 409 to the board Trello board pep. added PR 409 to the board Trello board

forgot the past tense

oh

jonas’ it's added

ok

sorry

I didn't, probably somebody did for me

I did

cool thanks

thx to you for raising it

I'm also interested about DOAP

If you can add that one as well :)

At least know who's responsability that is

well... not sure

Can we put this on the agenda so it gets decided who's responsible to say yes or no?

Also I got myself an account, pep@bouah.net

Ultimately, everything outside a XEP is Board.

pep. I have add_ED_ the item to the board's board

Ah ok thank you :)

dwd : and the board does not know/want, it's for the iteam

dwd: Well, mostly. Technical is still the domain of Council, even when non-XEP. E.g. technical direction. But mostly.

Do I need to be invited to the trello board? Should I be able to write stuff already?

Trello is fighting me so that I create my first board atm, I don't think I need that

ralphm, I have read, and carefully considered, your email to xmpp@ietf.

ralphm, It raises many interesting challenges.

ralphm, My response is as follows:

ralphm, That's a bit shit isn't it?

Everything is terrible. News at 11.

dwd: quite

ralphm, Didn't we discuss having a domain fully responsible for its own canonicalization at one point?

Like email?

ralphm, I've a vague recollection that Kurt Zeilenga suggested it could be used to canonicalize "Western Names" for Chinese people into the proper form, etc.

Zash, Like email but more so. We'd need a canonicalization service, I imagine.

dwd [20:35]: > ralphm, Didn't we discuss having a domain fully responsible for its own canonicalization at one point? What if that domain is malicious and sends major violations to us? Not looking at conference.jabber.org at all

Ge0rG, hand it to the SPAM WG?

dwd: maybe there's a sane strategy, but I'm unaware of one and it surely isn't documented clearly anywhere.

jonas’: no, what I mean is: what if it crashes clients or causes nasal demons?

that’s client bugs

Right, it's presumptuous to assume that the server won't give us JIDs with quotes in them

yes

guys, does https://github.com/dino/dino/commit/e84f2c49567e86d2a261ea264d65c4adc549c930 mean that 'from' always needs to be there?

jubalh, it depends on your stream layer

on the wire, the @from may also be omitted

because we have: https://github.com/profanity-im/profanity/blob/master/src/xmpp/roster.c#L215 right now

where we only test if equals barejid if from is there

that’s correct

if @from is absent, it’s to be treated as equivalent to your account bare JID

RFC 6120 something

(on inbound stanzas, obviously)

alright, thanks!

3. When the server generates a stanza from the server for delivery to the client on behalf of the account of the connected client (e.g., in the context of data storage services provided by the server on behalf of the client), the stanza MUST either (a) not include a 'from' attribute or (b) include a 'from' attribute whose value is the account's bare JID (<localpart@domainpart>).

some client libraries will alias absent @from to the bare JID for easier handling in client code

I tried removing redundant @to/@from at the server. IIRC some clients didn't appreciate it.

jubalh: some servers will do crazy things on messages inside of Carbons, at least for messages to your own JID

Yes that check in Dino has the potential to go boom. Because null is valid

But most servers do set it

I wish that'd be something the underlying xmpp would take care of

So it's fine ™

jonas’: the thing I was talking about with dwd is a bit more involved though, as per that email

Daniel: send a message to self on prosody 0.10 IIRC

Even if server do saner things, you still have issues with differing Unicode versions.

s

Myeah, prosody internally removes @to on stanzas to the account.

Ge0rG: yes I'm just saying that this particular check in Dino (around carbons) will be fine most of the time

But yeah if you do it properly you also want to accept null

Daniel, you missed the part where dino will automatically inject the correct value of from if it's messing

Ok

jonas’, can I somehow convince you to implement XEP-0392 in Profanity? ;) I heard you did it in poezio

So I have a JID getter wrapper that gets passed the appropriate default value for where it is called on.

https://github.com/dino/dino/blob/master/xmpp-vala/src/module/stanza.vala#L19

It even has a comment with reference to the RFC \o/

ralphm, "that email"?

https://mailarchive.ietf.org/arch/msg/xmpp/a-WhzOTyOq168GujQHgzQ1-DURI ?

right, I thought there was another email I was missing about solutions to the mess

jonas’, the one I'm going to write about potato farming? 😉

Zash, that sounds like a plan

No 🥔 no 🍟. Yay!

Moar 🥔

Thread relevant for retractions: https://www.openwall.com/lists/oss-security/2019/09/09/2

Brilliant. ß is mapped to ss by stringprep. Which is fine with IDNA 2003 because it does the same. However, in IDNA 2008 those are distinct.

what about ẞ?

Verboten!

nameprep (configured to allow unassigned) just passes it trough

IDNA 2003 doesn't like it IDNA 2008 maps it to 'ss'

$ lua -e 'print(require"util.encodings".idna.to_ascii("ẞß"))' xn--ss-hia

ẞash.

well mapping ẞ to what ever ß maps to makes sense. mapping it to ſs would also be an option I guess

but

ß maps to ß

someone should probably have a talk with them

about our beloved s'es

I vaguely recall reading about the same issue with domain names right?

idna2008 has a compat mode because of that

It does

more or less specifically because of the ß->ss issue

Yeah, but the non-compat mode seems to be moot due to stringprep already doing ß→ss

allowing anything but ascii was a mistake

Incidentally, I haven't spotted any non-ascii domains as active peers

So at least for domains maybe we don't have to worry too much.

someone on the Conversations had plä.net (or something) once

which triggered multiple things in Conversations

including cert validation and stuff

There's nödåtgärd.se which was around when Prosody originally gained IDNA code, and maybe even the reason for it.

Zash, maybe that's because they can't initiate anything with you that you haven't noticed them :P

> allowing anything but ascii was a mistake

haha yes

But punycode