XSF Discussion - 2023-06-13

I don't know any other messenger offering that. Reply is not the same as quote and not a direct replacement of it.

When you send emails, you're always replying to the whole email. You might be only quoting a part of the email, but the reply is always to the full email. And emails are far more likely to contain two independent topics that need separate replies than instant messaging.

So how do I reply to a message and quote a certain passage?

My idea would be to simply drop the fallback body

Only then the XEP is really what it says in the preamble, a improvement over quoting in attaching more metadata

Hm we can simply add quotes to a reply just will look funny in old clients

For some reason I was under the impression a reply does not allow quoting

If you add quotes in a reply it's likely to render semi-duplicated in clients, but of course there is not specified UI that's just what I've seen people doing so far

Maybe worth adding a 'quote' boolean to influence the UI?

One could do things like "if there is a reply fallback in the body assume we can show the quote"

Or users just get used to what their client does and compensate their workflow per usual, I dunno

I say 80% of users only ever do full quote, but there's a larger than expected number doing advanced stuff

i dont need a boolean, i would not change my UI if there are quotes in the reply

i think its not hard to make a UI that makes a distinct difference between what message the reply is to, and a quote inside the reply

lovetox, so I assume that means you think the quote should always be in the payload in full if it's a full quote?

and a client should never display the original message automatically in the case of a reply

not sure i understand you correctly, current XEP-461 sepecifies that the whole content of the replied message is in the body as a fallback

thats not optional

Ignore fallbacks for now, I'm talking about two clients which support XEP-0461

I would expect a client that supports replies, to render in some way the message which was replied to

but not in the form that a user would think its a "quote"

Didn't we have a references xep doing pointers to substrings? ✏

lovetox, well, in many popular messaging apps that's exactly how it's implemented

https://iphone-tricks.com/wp-content/uploads/2017/06/reply-directly-fast-whatsapp-message-3.jpg

yes i think this was understood, the question was if in the replies now quotes of parts of the replied message are allowed, and how they are rendered

Other messaging apps control the code on the sending and receiving side. In XMPP we don't, and what we don't want is client A sending a reply with an embedded quote, and client B rendering replies with a full quote above the body (as in the WhatsApp screenshot)

Yes, and the suggestion was to just use normal quote markup for that

yes

But if a client renders the full quoted message anyway, that will be weird

how is it weird? why assume messages are *always* rendererd in full

i would only plan to render one or two lines of the original message ✏

of course the user can jump to the message to read in full

It's weird because user of client B will see the full quote, followed by duplicate partial quotes of the same message

Okay, with your proposal they see many partial quotes of a message :)

only if the sending user wants to comment on individual parts?

Yes

should i disallow a user to respond to indivdual sentences of a big message?

How would you do that?

They can always type '>' followed by some text

copy/paste or whatever

i dont want to do it, you said it looks weird when there are quotes in a reply

i thought you had a idea what to do against it or how to do it better

My earlier suggestion was that the sending client can indicate if the reply should display as a full quote or not

Based on what UI it presented to the sending user

im not a fan of these ideas that i saw a few times

If the sending client puts the quote (possibly partial) in the body, then it would tell the receiving client not to render a separate quote

where some sending user trys to determine how a UI should look like on a receiving client

1. the sending user does not care in almost all cases

2. the sending user has no idea where the message will be received

Its like the "inline" element in 0447

its perfectly useless, i would never even use this information as a receiving client

Ensuring clients agree on the semantics of messages would be the primary reason we are all here :)

i will not let *other* clients determine how my UI looks

I'm not telling you how your UI should look

especially in a decentralized env, where *everybody* can write a client

It's basically a flag that says "there is quote embedded in the body"

If you want to ignore that and render the quote twice, that's up to you

But it would look weird and confusing

?! i render what the user sends, and if he quotes 3 times the same sentence

then i render it 3 times

Sure. If they quote it 3 times, that's what they expect the recipient to see

If they quote it one time, they expect the recipient to see it one time

But you will render it two times

ok i think now i know what you are getting at

It all depends on whether "reply" is an implicit quote or not

no a full quote will only sent in the fallback body

only partial quotes will be added to the body

So you will render a full quote, and then partial quotes, and you're happy with that?

reply is a full quote in my opinion

how can i not be happy with that, if the user wants to add partial quotes how am i to remove them?

then the context of his text is lost

I'm not saying you should remove them. It would be better to remove the automatic full quote, so they can just quote what they want.

i think this is something, thats better discussed if you saw the UI

the full quote is necessary its in the standard its called the fallback body

So yeah the full quote will always be in the body

and afterwards, follow partial quotes if the user wants

At no point have I been talking about the fallback case

I prefer to focus on what we want first, and then get a nice fallback after

yes attaching the reply servs as full quote

no need to add it to the body again (ignoring the fallback case)

So there is no way to do *only* a partial quote

and we're back to the start :)

i dont see it like that

Unless... > yes attaching the reply servs as full quote we change this ^

the problem is that you see a reply and a full quote as equivalent UI wise

but as i said in the beginning thats not a must

i plan to let a reply look different than a quote

its only a reference, and maybe i display part of that reference, but the user will know this is only a reference

while a quote will look different and the user will be aware here someone quotes a specific part of a sentence

and not just gives a reference so i know the context of his reply

i agree the difference is subtle, but i atleast will try to make that subtle difference evident in the UI

of course it will be more evident, when you reply to bigger messages, which i definitly will not show in "full"

and yes this maybe is only a thing for desktop clients

on smartphones this is all the same, as nobody writes long messages anyway

or starts quoting specific parts of sentences

its all to cumbersome on a phone, *if* even possible

> quoting specific parts of senten … look i just did 😁

> on smartphones Lol everything is a phone now isn't it? People run gajim on their phones

looks like XSF wesite builds on Github hang and don't complete

larma, regarding XEP-0444 Reactions, why did you choose for direct conversations origin-id or message-id as reaction-id. We could simply use stanza-id there too or not? We react to messages we receive, and stanza-id is always injected by the server.

Alex, thanks, should be ok now (just a hiccup)