XSF Discussion - 2020-10-02

does anyone still happen to have a link to the barbecue vs bazaar blog post (or what ever the analogy was) that argues that Slack and WhatsApp can’t be done in the same app

Daniel, https://blogs.gnome.org/tbernard/2018/05/16/banquets-and-barbecues/

wurstsalat, thanks

so it wasn’t bbqs and bazaars. no wonder I didn’t find anything with Google :-) ✏

docs.modernxmpp.org had it in its footnotes :)

i still find the analogy odd because it doesn’t really make sense to me which of those has more people

or is more intimate or whatever

Daniel: see also https://docs.modernxmpp.org/rationale/#group-chats

A barbecue is usually outside and informal, so random people from nearby can join and a banquette is invitation only. So banquette is private group and barbecue is public channel?

only that it is the exact opposite in the text

but yes that's why i'm confused

The terminology didn't age very well in the time of a global pandemic.

> only that it is the exact opposite in the text > but yes that's why i'm confused He, maybe BBQs are different at their place. Here you do them in the park or on the river and the BBQ groups happily mix and get together. Like on a festival. Banquette, at least for me, has the formal, invitation only, exclusive touch. 🤔

I have no clue what the context here is, but in my locale, a banquette does not imply any invitation type. It's more of a 'self-service' type of dinner setting, rather than you being served by a waiter.

Ah, that's what we call buffet here.

In their definition I think it's more of fundraiser

People don't know each other ✏

And are united by a single topic / cause

Yea, we frequently use 'buffet' too. Maybe I got my wires crossed.

Dict says > banquet (a large celebratory meal; a feast) (noun) Festmahl {n}, Bankett {n}

I don't think I've been to a barbeque of more than ~8 people

> I don't think I've been to a barbeque of more than ~8 people Go to the isar beach in summer. Hundreds if not thousands do barbecue. 😃

Probably says more about me than how the food is cooked

But I guess we can agree that those terms are not good chosen as it seems they get different interpretations.

Appears so

Has anyone else wanted to add 'favouriting' chats to their client, and if so what did they do? I'm thinking that a clone of the BM2 node but that contains favourites instead of bookmarks would be sensible for simplicity, although clearly BM2 could also be used with the <extensions/>.

Kev: Conversations has it

It's a shame that BMs are our mechanism for autojoining rooms, because logically what I want is to BM these chats (but don't want all my autojoined/known rooms to be BMd!).

Zash: Ta. Do you know what they're doing? Daniel?

I don't know more than that it pinns the chat to the top of the list of conversations

It a local feature

But publishing it will obviously be very easy once a XEP exists

I've wondered if maybe one of these decades we should go ahead and make "bookmarks" a native concept akin to rosters, possibly merge with rosters, since it's such a big part of chats.

Mental note for XMPP 2.0

Daniel: The per-chat notification setting would be valuable to the server for CSI and push notification decisions, fwiw.

That's separate, right?

Currently. Yes

But yes combining open and favorites might be a good idea

Zash: Rosters used to be perfectly usable for this, and allowed arbitrary extensions, and people habitually stored groupchats in them and stuff. But there was much wailing and gnashing of teeth and now people believe they should only contain Real People. Sadly.

Otherwise you might end up in weird situations where something is favorite but not open

Daniel: If I was to do something like using BM2 (although changing from BM1:private to BM2 would make this much less trivial for me) and store <favourite xmlns=.../> inside <extensions/> would that be something Conversations would adopt, or do you see issues with it? ✏

I think so. I'm not sure I'm ready to sync open conversations yet. Decoupling this would make it easier to implemt short term. But mid term coupling is the right way (see arguments above)

It's not just about 'open' conversations for me. I'd like a user to be able to 'pin' certain chats for easy access (or possibly even for the client to pin them automatically based on frequent use).

Right. But pinned is always a subset of open, no?

I'm not exactly sure what 'open' means.

pinned = intersection ( open, favorite ) maybe?

Ah, no, it's not.

Because 'open' only applies to MUCs, right? I'm interested in all chats.

So for MUCs Pinned is a subset of Joined, but there are also Pinned 1:1s.

Kev: When was that (anything in rosters)? Pre RFC 3921?

Right. I guess not all clients have the concept of open

Even up to early 3921bis times.

But in Conversations it means available in the overview not just in the roster

Or open tab in some clients I guess

Gotcha. Yeah, the last released version of Swift had tabs that probably correspond to this.

So in that case pinned *wouldn't* be a subset of open.

wait, using the roster for groupchats sounds ideal

you even get roster groups

until you consider the effects of sending presence

Yes. There are countless upsides.

and the confusion if a client doesn’t know it’s a group chat

jonas': There's nothing that says you have to send presence to things in your roster.

(Unless there's an appropriate presence sub, naturally)

If the server is fully aware that those are groupchats, then it can prevent you from trying too

I just got a very nice random email from someone that says "XEP-0418: In the schema the base64Binary should be xs:base64Binary." does anyone know if that is true? https://xmpp.org/extensions/xep-0418.html#schema ✏

I'd say "probably"

Yes, the xs prefix being already declared above

looks like I need to update this text anyway: "XEP-XXXX: https://xmpp.org/extensions/inbox/dox.html"

heh

so it *should* be prefixed by xs: ?

Is any other type in XEPs prefixed with xs? Does anybody know how to Schema at the XSF?

the way you currently do it it's pointing to an undefined type

I certainly do not (know how to Schema)

put the schema and the xml in a schema validator and see if it validates 😉

ok well I picked 4 other random XEPs and all of them have "schema will be supplied at some later point" LOL

I did a thing which lints xeps including schema validation.

I get the impression I should clean it up

https://xmpp.org/extensions/xep-0065.html#schema these all say type='xs:...'

I'll put in a PR fixing both of those issues and let council sort it out :)

moparisthebest, according to https://www.freeformatter.com/xml-validator-xsd.html you need to put xs:base64Binary. Also your example "vOIBIAABAAAAAAABB2V4YW1wbGUDb3JnAAABAAEAACkQAAAAAAAADAAKAAj5HO5JuEe+mA" is actually invalid base64, it must be "vOIBIAABAAAAAAABB2V4YW1wbGUDb3JnAAABAAEAACkQAAAAAAAADAAKAAj5HO5JuEe+mA==" 😉

moparisthebest, yeah schemas in XEPs are a mess :(

larma, "In both the query and response, the content of the <dns/> element is the DNS on-the-wire format as defined in RFC 1035 [1]. The body MUST be encoded with base64 RFC 4648 [5]. Padding characters for base64 MUST NOT be included." :)

Or well, they aren't, because they don't exist

moparisthebest, well in that case you are not using the xs:base64Binary type

maybe... where is *that* defined

https://www.w3.org/TR/xmlschema-2/#base64Binary

> Note that this grammar requires the number of non-whitespace characters in the lexical form to be a multiple of four, and for equals signs to appear only at the end of the lexical form; strings which do not meet these constraints are not legal lexical forms of base64Binary because they cannot successfully be decoded by base64 decoders.

well that's annoying

Write your own type! :P

:/

just joking, but it's pretty tempting :)

the reason I forbade padding characters was because I figured this would often be used as a proxy or substitute for DoH and it does the same https://tools.ietf.org/html/draft-hoffman-dns-over-https-01#section-5.1

actually no, it specifies base64url and I specify base64, how unfortunate

no memory if that was a simple mistake or I did it so I could use the base64Binary schema type...