XSF Discussion - 2019-03-08

Wiktor, anyone else interested, code is up https://github.com/moparisthebest/jDnsProxy/blob/dox/dox.md

also I have it running at dns@moparisthebest.com/listener if anyone wants to query some DNS over XMPP

“20:04:36 flow> I never understood why we use numbers when humans are much better with letters”, Western* humans, just ask a random Chinese person to remember letters instead of numbers and you’ll have a bad time.

Link Mauve: not even if those letters are abbreviations of the subject matter name?

AIUI, even then.

Link Mauve, TIL. But is that an argument that we shouldn't use short names instead of numbers?

But, we do have short names

that nobody uses

that makes the situation even more ironic, we have them, I hope they are supposed to be short and unique, and still we use additinally numbers

But IIRC they get assigned at Draft or something

> wurstsalat: it is already? pep., I meant on the xmpp.org frontpage, with a banner?

Ah you want the fancy banner :p

If you come up with the CSS for it I assume somebody will accept it

pep., still no computer access for me until end of march, sorry. next time it will be a PR ;)

flow, the numbers can be uniquely searched for, the short names may be so short that they occur in other words

so using the numbers is much better for searchability

at least that’s my rationale for using them in the ML subject and in rooms

jonas’, XEP-SM?

then we can never have XEP-SMART

XEP-CSI, XEP-CARBONS, XEP-FMUC

and what about avatars

or bookmarks

I immeditaly will know what XEP-FMUC is about, can you tell me the FMUC number from the top of your head?

XEP-BOOKMARKS XEP-BOOKMARKS-DONE-RIGHT ...

no, since I never cared about FMUC

jonas’, right, but it doesn't make the situation worser

the others I use I typically know, but I’m a number person

also XEP-CAPS, and XEP-CAPS2 works pretty well for me

then you can’t search for the old caps only

you’ll always get both

I think I could live with that

not much more frustrating than having inaccurate search results and not being able to do anything about it

it would be even better if we could tag threads and messages of the ML with unique xep tags for searching, but that is a different discussion

I agree

> Western* humans, just ask a random Chinese person to remember letters instead of numbers and you’ll have a bad time who cares, there is already a ton of language discrimination 🙂 It's even more sharp when your native alphabet is not latin

this stringprep crap for example, not every language has a problem with character cases, and this shit is built into the protocol, permanently consuming resources at the servers

ZİNİD!

The obvious solution is to use XEP Emoji Shortnames.

I would like to propose a minor change to XEP<FacePalm><Male><Fitzpatrick-Modifier-3><ZeroWidthJoin><Regional-Indicator-Letter-Ф>

User was banned for this crime against sanity

+1

FWIW, I think we often do refer to protocols by name, and not number. We refer to the documents by number for precision, but there's only a handful that are routinely referred to by number (198, for example). Most of the time we talk about MUC, or vCards, or whatever.

does "MUC" also include all the undocumented quirks that are not part of 0045? 😁

Ge0rG, fancy documenting those?

What undocumented quirks are there? I can think of the vCard hack and MSN

MattJ: MSN is the most prominent example. I think *somebody* started a wiki page on it, but then it was widely ignored by the people who know the most about their respective implementations.

Was that page among the losses in the great server crash? ✏

Zash: no

At least I don't think so

I think I started writing a sentence about the vcard hack but git ate it, probably when I looked at GC 1.0 removal

Hah, https://xmpp.org/extensions/xep-0089.html

What was the conclusion of looking at an old XEP each week?

Zash, Everyone thought that someone else should definitely do that?

Let's talk about XEP-XXXX: Tragedy of the Commons

Have we changed to roman numerals?

What do people think about this: https://lab.louiz.org/poezio/poezio/issues/3462 ? poezio bug: "LMC replaces the whole stanza"

pep. given that we're already in a situation where both approaches exist, an implementation note that warns to expect both methods might be good.

Replacing the whole stanza isn't what 308 asks you to do.

Just the payloads.

that's getting rather philosophical.

Stanza of Theseus

client developers problems

does poezio save stanzas to its local storage?

If you are a XSF member and have not voted yet in our current voting period then please do so. Memberbot is online and waiting to accept your votes ;-) Thank You

Done, thanks :)

(Y)

The deadline is 2019-03-12 19:00 UTC!

Alex: thanks for running all this process 👍

done, thanks!

Alex, very appreciated :)

New applicants :O

Yeah!

lovetox, no it doesn't

then how does it "replace" a stanza

What do you call local storage?

are you saying it keeps the received stanzas in some cache?

yeah, there's a buffer of stanzas that get displayed or sth

(I don't exactly know the details yet)

LMC will go as many times as it's corrected in the logs in any case, I think

But not displayed

(but **poezio will not display all of them, unless you ask for it)

yes, but you wrote it replaces a stanza, but now from what you tell it does sound like it does put every message into storage

By replacing the whole stanza we also lose the ability to see corrections _after_. You have to have that setting _before_ that happens

so i dont see replacing going oon

lovetox, there's hot/cold storage :p

what about warm storage?

lovetox, when I say "replace" in that issue, I mean that the original stanza is getting replaced by the correction, in the buffer that will get displayed

(I think that's how it works, Link Mauve/mathieui will know better)

not the stanza but the displayed text

no, the stanza

(I'm pretty sure that LMC doesn't apply in the xml tab)

Yeah that's a different tab..

But we keep the stanza, with the associated ids/attributes etc., and that's actually what's causing the issue here, because we don't know about the original ID anymore

So we expect other implementations to refer to the one-to-last id

Oh, I should have read the issue first.

:)

pep. but every impl refers to the last message

not to the first

if i remember correctly

lovetox, not every no

converse doesn't

And the XEP is not explicit about it

then you have no base for action

so why change anything

everything?

hm no i said anything :D

Ge0rG, I like your time machine

ah, anything.

tl;dr: nobody cares.

lovetox, well, because some LMC messages won't appear as LMC

And it's still valid

you mean you impl displaying both ways?

pep.: tell converse that it's called LMC and not FMC for a reason.

Ge0rG, we could do that by being explicit in the XEP yes

Or we could fix poezio to support all cases

lovetox: yaxim only stores _one_ message in the DB (yeah, I know, I know). So it can only follow a linear LMC trace and not a tree.

Ge0rG, Gajim stores all messages but still corrects the last

technically, it should suffice to store two IDs, the _first_ ID and the _last_ ID for correction purposes.

but then you need to index on both.

lovetox, yeah but at least you know the id converse is referring to. poezio doesn't

i know the id but i still dont show the correction

heh, why?

because i only allow the last message to be corrected

pep.: I suggest you write a PR outlining that always the lastest message is to be referenced.

hmm

its called last message correction :D

lovetox, It's not called "last <message> correction" though

i dont see the difference

ah i get it

you mean the user decides what counts as a message

XEP-XXXX: Message Correction Like Last Message Correction, but any recent message, within reason.

There I fixed it

lovetox: I also have a time machine deployment for you: https://mail.jabber.org/pipermail/standards/2018-June/035143.html ;)

Ge0rG, I want access to your time machine

either way this convers stuff is a good clue

<feature var="any-message-correction"/>

because i have a bug report with not working LMC

and now i think i know why

pep.: it's called mutt <esc>/~S

hehe

lovetox: to quote from my past self: > I think there is a compelling reason to allow correcting more than just the last message - imagine typing multiple lines in a row, and only then reading what you sent, to realize a typo / incorrectness.

im not against it within reason

Mattermost allows Any of My Messages Correction, unless you're an admin and then it's YMC

Any makes not much sense to me, we are chatting here, this is not a document

Tell that to users with strong OCDs

also its too much impl work

lovetox: yaxim defines "within reason" as "the last 10 stored messages from that user"

because when people can alter your history next thing your users want it to be made aware of it

lovetox, I agree

as corrections are overwriting the original message, you can do an unlimited number of corrections of the same msg.

or as /marquee

MattJ: showcase!

Here? In this MUC?? ;)

In this part of XMPP?

MattJ: dunno. Record a screencap maybe?

Alas, it's a private project, and I don't intend to make this public because someone worse than me will use it somewhere

Ge0rG: Don't give them ideas!

[=====> ] 23.42KB/s ETA: tomorrow

Not when /marquee is a thing already

is there a way to determine the width of the recipient's terminal over XMPP?

Put it in disco#info

XEP-XXXX: RTW

<winch columns="80" lines="24"/>

Ge0rG, do you send something different for each participant?

Or is that a server thing?

pep.: yes!

<width columns="273" lines="74"/>

SIGWINCH

in PEP

https://op-co.de/tmp/progress.mp4

I did it.

Now let's make 2D animated art.

I don't know if I'm impressed, or disappointed.

can't you be both, along different aspects of it?

Star Wars over XMPP

(now somebody must make the stupid name joke)

Wow, it's still up

Pft Star Wars