[Standards] Proposed XMPP Extension: Quick Response

Marvin W xmpp at larma.de
Tue Apr 21 12:23:39 UTC 2020


To me that still sounds like the only major difference is that actions
are not backwards compatible.

I understand that action IDs can be just random, however I wonder what
the usecase of such is that couldn't be realized using a human-readable

Looking at the example in the XEP, the action id = response value of a
an action/response labeled "Merge now" for a merge request with id 3
could just be "Merge MR!3" which is both human and bot readable. I think
something is wrong with an action if it cannot be described in
human-readable terms.

With regards to fallback, I was thinking that I'd like the fact that I
performed an action be visible on other clients (through MAM/carbons)
even if they don't support this new XEP. For Responses this works but
for Actions it doesn't.


On 4/21/20 1:18 PM, syndace at web.de wrote:
> Hi Marvin,
> I'll try to summarize the difference between Responses and Actions:
> A Response is supposed to be exactly equivalent to a "manual" text
> response, without any magic implied. That is, manually typing "yes" in a
> client that doesn't support Quick Response is 100% equivalent to
> selecting the "yes" Quick Response in a client that supports the XEP. An
> implication of the absence of "magic" here is, that Responses only apply
> to the most recently received message, because without magic there is no
> "reference" from the response to the original message.
> Actions are not compatible with clients that don't support Quick
> Response, as they require "understanding" the <action> element and
> responding with an <action-selected>. Because clients are free to
> generate unique ids here, the selection of an action is not limited to
> the last message, there is no problem relating the response to the
> original message. A "fallback" body is something that bots can choose to
> do anyway, by allowing to trigger the action via a plaintext response
> too (or comparable).
> Hope that clears it up,
> Syndace

More information about the Standards mailing list