- Standards

XEP-0115: Mismatched 'node' attribute value when detecting capabilities

by Guus der Kinderen

Hi all, This follows a discussion the XSF's Standards Discussion MUC. Logs start at https://logs.xmpp.org/xsf/2025-05-18 (and continue into the next days). XEP-0115: Entity Capabilities describes how capabilities from the "generating entity" can be discovered by a "requesting entity" in section 6.2. To do this, the "requesting entity" sends a disco#info request (that MUST include a disco 'node' attribute for backwards-compatibility). The 'node' attribute value is generated in a way that results in an application-specific value that will be specific to the application that generated the string as well as the capabilities that are currently advertised by that application. What the XEP does not describe, is what is to happen when the requesting entity uses in their request a 'node' attribute value that is different from what the generating entity would produce (given its current configuration). Since providing the value, the generating entity could have undergone configuration changes making available a different set of capabilities, thus making the original value outdated. The XEP is improved by explicitly describing this scenario, by defining what should happen when a requesting entity uses a conflicting 'node' attribute value. I can think of a number of options: 1. If the 'node' attribute value sent by the requesting entity is recognized by the generating entity as a valid value in some context (eg: a value that represents an earlier state of offered capabilities), the generating entity could respond with the corresponding capabilities. Downsides: the requesting entity may take this as an indication that this represents the _current_ capabilities of the requesting entity, when that may not be the case. This approach can't be used if the generating entity does not recognize the value 2. The request could be answered with an error (eg: item-not-found). Downside: the requesting entity, that only performs this request because it's interested in the current capabilities of the generating entity, needs to do another round trip to get the answers it is looking for. 3. The request could be answered with a response in which the generating entity uses the new/up-to-date 'node' attribute value (it explicitly does not echo back the value that was in the request). Downside: there is a potential backwards compatibility issue for existing implementations that currently use another approach. They may end up mismatching 'node' attribute values and associated capabilities. Given the wording in section 6.2, specifically the definition that the requesting entity is to include the 'node' attribute for backwards compatibility, seems to suggest that the responding entity may not have a need for this value. That, to me, suggests that option 3 is most appropriate. I suggest that section 6.2 of XEP-0115 is modified to include an explicit description of option 3. The backwards-incompatibility issues are arguably existing only because of unspecified behavior. I do not see how such issues would be significantly affecting functionality. As the XEP is currently in state Stable, backwards incompatible changes are allowable. Thoughts? Kind regards, Guus

8 months, 3 weeks

1
0
0 0

mailing list move

by Peter Saint-Andre

We're in the midst of moving this email discussion list from one server to another. You might be asked to re-confirm your subscription. Peter

9 months, 1 week

1
0
0 0

Threads and replies in XMPP

by Goffi

Hello, We've got a discussion about thread on xsf@ a couple of days ago, and I would like to bring it here. I initially thought that XEP-0461 (Message Replies) was redundant with threads, except in the case where there is no thread ID in the message a user wants to reply too. After discussion, notably with singpolyma and lovetox, it appears that we don't have the same view on this. The way I got it, we have basically 2 ways to see threads: 1. A series of replies to any message in a group chat, à la Slack (in Slack, you see the replies hidden under parent message, with something like "X replies", and when you click, a panel appear with the whole thread. 2. A side discussion explicitly started, that could be seen like a new chat window (UI example: user clicks a "start a new thread button", and start a blank window with it message, this message has a new thread ID, that will be used by people replying to it). In case 1, the parent message needs a thread ID so that thread can continue with the same ID. Problem is if parent message doesn't has a thread ID (because client doesn't support thread, or because there is no thread ID for each message), in this case a XEP-0461 "reply to" can be used, then the first reply has a new thread ID, and the thread can continue. In case 2, the explicitly started thread has a newly generated thread ID, so there is no problem. I want to support both 1 and 2. For 1, my initial reading of XEP-0201 (Best Practices for Message Threads) was that a new thread ID should be generated for each message: > Unless a <message/> stanza is written in direct reply to another <message/> > stanza, if a ThreadID is included then its value SHOULD be newly generated > when a human user initiates a chat conversation with another user (i.e., a > <message/> stanza of type 'chat'), starts a new conversation in the context > of a multi-user chat environment (i.e., a <message/> stanza of type > 'groupchat'), or sends a normal message. XEP-0201 §3.2. But after discussion, I realise that it's not requested to generate a new ID each time, and it seems that generating new thread ID for each message would be a problem for some clients (notably Cheogram which would create a thread UI for each ID). The solution would be to use "reply to" from XEP-0461, then a newly generated thread ID. I've also proposed an alternative when parent message ID could be used as thread ID, this may we can find parent message without relying on XEP-0461. For the moment, I think that I won't generate new thread ID for each message, and if I create a thread from a parent message I'll use a XEP-0461 "reply to" and generate a new thread ID, then use this thread ID for following message in this thread. However, I would like to have feedback from the wider community: - how to you handle or plan to handle threads? - Are you planning to implement case 1 (threads can be started from any message) and/or case 2 (threads are explicitly started)? - How would you handle case 1? Thanks! Best, Goffi

9 months, 2 weeks

5
8
0 0

A proposal to update XEP-0372: References

by Schimon Jehudah

Good day. Last week, I have contacted various of projects and asked to render messages with XMPP URI by their particular context. Please refer to the Gemini page or see the attached file for references. gemini://woodpeckersnest.space/~schapps/journal/2025-04-24-rendering-xmpp-uri-by-context.gmi After communicating this information to nephele, the developer of Renga chat client for Haiku, he has advised to standartize this approach by adding a new attribute to XEP-0372. Sun 27 Apr 2025 06:03:34 PM - nephele: > The XEP 0372 as it exists now seems to almost hit what you want, maybe > only another type is needed. for example <reference xmlns='urn:xmpp:reference:0' begin='72' end='78' type='adhoc' uri='xmpp:nephele@xmpp.org?command'/> Please advise. Kind regards, Schimon

9 months, 2 weeks

2
1
0 0

XEP-0424/0425/0461: Spec misses guidance which ID should be used in case of corrections

by Philipp Hörist

Hi, All these XEPs miss a paragraph how corrections should be handled correctly e.g. if the ID of the original message or of the correction should be used. This leads already to different implementations in popular clients, none of which can be considered wrong as the XEPs do not specify a correct handling. Regards Philipp

9 months, 2 weeks

3
8
0 0

URI Query Types for Ad-Hoc Commands and Data Forms

by Schimon Jehudah

Good day. I am the developer of Slixfeed news bot. Data Forms are extensively utilized by that bot. I think that it would be beneficial to define IRI/URI to acess to Ad-Hoc Commands and Data Forms. == Ad-Hoc Commands == This IRI/URI would open the form "manual". xmpp:slixfeed@xmpp.org?command=help This IRI/URI would open the form "about". xmpp:slixfeed@xmpp.org?command=about == Data Forms == Provided that a Data Form has an identifier, this IRI/URI would open the about dialog for the given query. This IRI/URI would open the form "about" (same as of the example of the Ad-Hoc Command, which can lead to the same form by explicitly define it). xmpp:slixfeed@xmpp.org?form=about-start This IRI/URI would open the form "about" in the next form to form about-start with the value "thanks". xmpp:slixfeed@xmpp.org?form=about-result?payload=thanks Please advise. Kind regards, Schimon

9 months, 2 weeks

1
1
0 0

A question concerning to XEP-0128: Service Discovery Extensions

by Schimon Jehudah

Greetings. Is it possible to restrict the set of reactions (XEP-0444) to messages which a *client* entity of type *bot* sends to private and public chat? Regards, Schimon

9 months, 3 weeks

1
0
0 0

UPDATED: XEP-0455 (Service Outage Status)

by Daniel Gultsch

Version 0.3.1 of XEP-0455 (Service Outage Status) has been released. Abstract: This document defines an XMPP protocol extension that enables server administrators to communicate issues with the server to all users in a semantic manner. Changelog: Fix typo in JSON Schema. (ka) URL: https://xmpp.org/extensions/xep-0455.html Note: The information in the XEP list at https://xmpp.org/extensions/ is updated by a separate automated process and may be stale at the time this email is sent. The XEP documents linked herein are up-to- date.

9 months, 3 weeks

1
0
0 0

XEP-0433 Extended Channel Search feedback

by Guus der Kinderen

Hello! Here's my feedback on XEP-0433 Extended Channel Search: https://xmpp.org/extensions/xep-0433.html The feedback herein is based on two efforts: - Adding support for XEP-0433 to Openfire (which is now reaching a ready-ish state: https://github.com/igniterealtime/Openfire/pull/2756 ) - Adding interop tests to the Xmpp Interop Testing Framework ( https://github.com/XMPP-Interop-Testing/smack-sint-server-extensions/pull/77 ) Particularly the exercise of trying to write tests generated quite a bit of feedback. Without having explicit MUST conditions, it is not possible to write many assertions. This is why a lot of the feedback below suggests making things explicit. 1. I'd like to see a MUST in 4.2 / 4.2.1 that describes that the service to be searched MUST support the retrieval of a search form. That removes ambiguity, I think. I prefer the clarity of the wording in the comparable section 2.1 of XEP-0055: https://xmpp.org/extensions/xep-0055.html#usecases-search That section describes basically the same functionality requesting a form), but it's more explicit. 2. Explicitly define the form type (which is now mostly in an example. It's slightly different from Muclumbus, which can add to the confusion (maybe a 'differences with precursor' section would be handy). 3. The XEP defines two sort keys. It would be good to explicitly define if they are mandatory-to-implement? 4. If the sort keys are meant to be extensible, like the form fields, it would be good to make that more explicit (perhaps with references to a registry, like we do for other XEPs). The XEPs author mentioned in a chat that sort keys are meant to be extensible and neither are mandatory-to-implement. They also wrote that clients must discover support for individual sort keys by requesting the form. No matter what detection mechanism is used, I would like to see a MUST definition for the server to provide this mechanism. Without it, clients can only guess. 5. Section 4.2.2 reads: "it MAY omit all fields it does not know or where it does not change the value already supplied by the Search Service." This implies that the Search Service will use the default value of that field, if not supplied by the Searcher. If that's the intent, I'd also explicitly state that, to prevent confusion. I do wonder if that can cause weird issues. A searcher can submit the exact same search request (to different services) but it could represent a very different query. I wonder if that's going to cause problems at some point. The XEPs author replied via chat: "regarding the defaulting: you're right: absent fields should be interpreted as "client does not know them" as opposed to "use defaults"." Let's change the XEP to reflect this. 6. Section 4.2.2 reads: "The Searcher MAY include a Result Set Management (XEP-0059) [4] <set/>" Does this imply that the Service MUST support RSM? If that's not the case, then I would change 'MAY' for 'may' here. As XMPP is extensible, any entity should disregard extensions that it doesn't recognize anyway. With that in mind, the MAY usage seems to imply something. If that _is_ the case, then I'd prefer that to be explicitly stated (eg: "the Service MUST support RSM"). The XEPs author replied via chat: "yes, the service MUST support RSM. RSM is a hard dependency. (both sides MUST support RSM in fact, but the initial query does not necessarily need an RSM element, in which case page size etc. are defaulted by the service)." Again, I propose to make this explicit in the XEP. 7. Question on XEP-0433 / 4.2.2.6 (conflicting fields): Would usage of `all` combined with one of the vars that relate to `q` (like `sinname`) also be considered conflict to which a `<conflicting-fields/>` MUST be responded with? Is using `sinname` (and/or friends) _without_ `q` (or any equivalent, unspecified var) also considered a conflict? The XEPs author replied via chat: "yeah, I think `all` should conflict with all other fields which may restrict the resultset. `sinname` without a query or equivalent should also cause a conflict, yes." I answered: "even `all` combined with something like `sinname=false` ? Things get tricky there :) (and is that a MUST NOT or SHOULD NOT?) (which is increasingly annoying when the form is pre-populated by the server, as not supplying a field does not change the value already supplied by the Search Service. Doesn't this mean that the server can never pre-populate `sinname` and friends, as you can't un-populate those when you want to do a `all`?)" 8. example 10 https://xmpp.org/extensions/xep-0433.html#example-10 uses `not-allowed` while the text defines that `bad-request` should be used. 9. Table 1 defines `sinaddress` while example 3 uses `sinaddr` - muclumbus used `sinaddr` which makes this an easy one to get wrong, I think. Perhaps another item for a 'differences with precursor' section. 10. The XEP has "ignoring text search terms" in a couple of places that relate to `all`. I would remove that, as it implies that any value for `q` could be ignored (rather than that this is a MUST conflict scenario). 11. Section 6.2 describes the 'is-open' as boolean character data. "If set to true, it indicates that the channel can be joined without extra credentials." In examples, the element is used like this: `<is-open/>`. This to me seems to suggest that the element is present when the room is open, which suggests that the element is absent when the room is not. This does not track well with the definition, which seems to call for an explicit 'true' or 'false' value. 12. Somewhat of a nit, but table 2 doesn't define what the direction of the ordering is when using the address-based key. Consider making that more explicit. That's it, I think. Apologies for the haphazardous approach. This was compiled in many different sittings, which made the organisation go out of the window. - Guus

9 months, 3 weeks

1
0
0 0

Hosting the XEP index over XMPP

by Schimon Jehudah

Good day. The XEP index is currently hosted over HTTP, and I think that it would be a good idea to host it over an XMPP server as well. Actually, until some time ago, I have used an HTML browser just to access XEP pages, when I did use a browser. I have also wrote an article about hosting sites over XMPP. gemini://woodpeckersnest.space/~schapps/journal/2024-12-14-xmpp-as-the-internet.gmi A video from Goffi. https://fosdem.org/2025/schedule/event/fosdem-2025-5721-a-universal-and-sta… I am also eager to for the day when Libervia standardize an XEP for forum, so that also discussions of XEP be feasible over XMPP with Libervia and Movim. Regards, Schimon

9 months, 3 weeks

2
1
0 0

2026

2025

2024

2023

Standards