Proposed XMPP Extension: Bookmark last known name
The XMPP Extensions Editor has received a proposal for a new XEP. Title: Bookmark last known name Abstract: This document defines an XMPP protocol extension to allow users to pin PEP Native Bookmarks. URL: https://xmpp.org/extensions/inbox/bookmark-last-known-name.html The Council will decide in the next two weeks whether to accept this proposal as an official XEP.
Big fan of these new proposals, quite excited to see where they go :) Just one minor note: this is the same Abstract text as Bookmark Pinning, which I assume wasn't intended. On 25 October 2026 5:28:25 am UTC, Daniel Gultsch <daniel@gultsch.de> wrote:
The XMPP Extensions Editor has received a proposal for a new XEP.
Title: Bookmark last known name Abstract: This document defines an XMPP protocol extension to allow users to pin PEP Native Bookmarks.
URL: https://xmpp.org/extensions/inbox/bookmark-last-known-name.html
The Council will decide in the next two weeks whether to accept this proposal as an official XEP. _______________________________________________ Standards mailing list -- standards@xmpp.org To unsubscribe send an email to standards-leave@xmpp.org
On Sun, 25 Oct 2026 at 05:44, Daniel Gultsch <daniel@gultsch.de> wrote:
The XMPP Extensions Editor has received a proposal for a new XEP.
Title: Bookmark last known name Abstract: This document defines an XMPP protocol extension to allow users to pin PEP Native Bookmarks.
URL: https://xmpp.org/extensions/inbox/bookmark-last-known-name.html
The Council will decide in the next two weeks whether to accept this proposal as an official XEP.
I don't object to this - it's harmless, and well designed as far as it goes - but I'm struggling to understand the rationale. So a client, which has no information cached, wants to show a list of bookmarks, but doesn't want to use the name the user explicitly chose, and presents the "official" name of the chatroom instead? Did I get that right? But only the official name when some other client that supports this protocol last checked? Obviously, clients can do whatever they want, but ... why? Why ignore what the user explicitly directed? If what we want to do is shortcut the disco#info for bookmarked rooms, BTW, we can absolutely do that. Imagine a bot-like client that preiodically republishes the disco#info to the item. And that could reasonably be built into the server. Anyway, all this might be just me missing the point, or not being hip with the yoof, or something. Dave.
Hi, Yes i think you miss the point, but there are some confusing things in the XEP. First i dont see where the XEP says, the client does not want to use a name the user sets. If the bookmark name attribute is filled, this should be used, and this XEP has no relevance in this case. I think the idea is a gateway like for Matrix, fills your Bookmarks with matrix chatrooms. Now as a client you log in, and get 50 known matrix rooms synced. But these are just "known" rooms, not necessarily rooms you want to autojoin on that client. If you dont join the room on sync, no client will disco a room which we dont intend to join. Hence we have nothing in the Bookmark but the JID. So this would be a way, how a gateway can tell you upfront what the name of the groupchat is, so you dont just have the JID. Now why does the XEP talk about the "name" attribute, because it tries to counter the argument that someone potentially raises, "Why dont you publish the name into the existing name attribute". I think this would be unclean, because the name attribute is currently filled only by users, if they assign a name themself. If a server sets this, a client cannot differentiate anymore between a user chosen nick and one another identity assigned. Hence the whole proposal for an additional attribute. This has also relevence in non-gateway cases. For example in Gajim we store all known MUCs in bookmarks, also ones we do not join anymore currently. If i setup Gajim on a new client, it downloads all the MUCs, but for the chats we dont autojoin, we only have the JID. It would be nice if a server component which has all disco info, can simply update my bookmark so it always has the right server name, without me executing disco on all known mucs. Regards Philipp On Sun, Oct 25, 2026, at 20:38, Dave Cridland wrote:
On Sun, 25 Oct 2026 at 05:44, Daniel Gultsch <daniel@gultsch.de> wrote:
The XMPP Extensions Editor has received a proposal for a new XEP.
Title: Bookmark last known name Abstract: This document defines an XMPP protocol extension to allow users to pin PEP Native Bookmarks.
URL: https://xmpp.org/extensions/inbox/bookmark-last-known-name.html
The Council will decide in the next two weeks whether to accept this proposal as an official XEP.
I don't object to this - it's harmless, and well designed as far as it goes - but I'm struggling to understand the rationale.
So a client, which has no information cached, wants to show a list of bookmarks, but doesn't want to use the name the user explicitly chose, and presents the "official" name of the chatroom instead? Did I get that right? But only the official name when some other client that supports this protocol last checked?
Obviously, clients can do whatever they want, but ... why? Why ignore what the user explicitly directed?
If what we want to do is shortcut the disco#info for bookmarked rooms, BTW, we can absolutely do that. Imagine a bot-like client that preiodically republishes the disco#info to the item. And that could reasonably be built into the server.
Anyway, all this might be just me missing the point, or not being hip with the yoof, or something.
Dave. _______________________________________________ Standards mailing list -- standards@xmpp.org To unsubscribe send an email to standards-leave@xmpp.org
On Sun, 25 Oct 2026 at 19:59, Philipp Hörist <philipp@hoerist.com> wrote:
Yes i think you miss the point, but there are some confusing things in the XEP.
Thanks, yes.
First i dont see where the XEP says, the client does not want to use a name the user sets. If the bookmark name attribute is filled, this should be used, and this XEP has no relevance in this case.
Right, this is the missing feature.
I think the idea is a gateway like for Matrix, fills your Bookmarks with matrix chatrooms. Now as a client you log in, and get 50 known matrix rooms synced. But these are just "known" rooms, not necessarily rooms you want to autojoin on that client.
If you dont join the room on sync, no client will disco a room which we dont intend to join. Hence we have nothing in the Bookmark but the JID.
So this would be a way, how a gateway can tell you upfront what the name of the groupchat is, so you dont just have the JID.
Got it. Now I'm wondering why a gateway would publish the name of all available groupchats as bookmarks. I do, however, run into some related problems from time to time. For example, joining a chatroom on my local server is easy - my client grabs disco and uses it for type-ahead search. But joining one on a remote server is hard because my client doesn't know which server I might want. Being able to have a bookmarked list of MUC services to disco would feel like it's solving this problem as well as mine. (Bearing in mind the name is in disco#items).
Now why does the XEP talk about the "name" attribute, because it tries to counter the argument that someone potentially raises, "Why dont you publish the name into the existing name attribute".
I think this would be unclean, because the name attribute is currently filled only by users, if they assign a name themself. If a server sets this, a client cannot differentiate anymore between a user chosen nick and one another identity assigned.
Hence the whole proposal for an additional attribute.
Right, this makes more sense now.
This has also relevence in non-gateway cases.
For example in Gajim we store all known MUCs in bookmarks, also ones we do not join anymore currently.
If i setup Gajim on a new client, it downloads all the MUCs, but for the chats we dont autojoin, we only have the JID. It would be nice if a server component which has all disco info, can simply update my bookmark so it always has the right server name, without me executing disco on all known mucs.
And with that, I'm now confused again. Easy to do. All known MUCs? Presumably this is all MUCs that Gajim has ever joined, or is it all MUCs it's ever discovered via disco, MUC search services, etc? Dave.
Hi
And with that, I'm now confused again. Easy to do.
All known MUCs? Presumably this is all MUCs that Gajim has ever joined, or is it all MUCs it's ever discovered via disco, MUC search services, etc?
Bookmarks are a way in XMPP for clients to hold a list of group chats it knows and wants to remember. What "knowing" means, is up to the clients. Usually this means, a group chat the user was a part of at least once. Such a concept i would assume is als present on other chat networks. If you login to a gateway for Matrix, you expect that your "Known Group Chat List" now also knows all the known Matrix Chats from your Matrix account. The most obvious way for a gateway to sync that info would be to add them to your bookmarks. (for contacts we have the roster where the same mechanism exists) Now you have a bunch of matrix rooms in your bookmark list, but no name for them, and no attribute for the server to put it. Gateways are a good example, but of course we can imagine that your homeserver does fill such a attribute for you also for XMPP MUCs. Regards Philipp
Hi, Philipp already defended this better, so I don't have much to add. A few small points though:
but doesn't want to use the name the user explicitly chose
My wording must be off somewhere because that is precisely *not* the intent. :) Slidge used to write in the basic 'name' field for bookmarks (roster items too btw) but this is a bad approach. These fields are meant for user-set custom names, which should not be overridden by anything automated.
we can imagine that your homeserver does fill such a attribute for you also for XMPP MUCs.
Indeed, a 3rd use-case to add, thanks.
Now I'm wondering why a gateway would publish the name of all available groupchats as bookmarks.
Because that is what clients use for the "start chat"-like dialogs so that's the best integration. FTR, slidge also provides an adhoc command to list the groupchats you can join. This is fine but not as well integrated from the end-user side.
This document defines an XMPP protocol extension to allow users to pin PEP Native Bookmarks.
Oh damn, got my hand caught in the ctrl+c/ctrl+v jar! Thanks for your feedback. -- Nicolas
On Sun, 25 Oct 2026 at 20:46, Philipp Hörist <philipp@hoerist.com> wrote:
Hi
And with that, I'm now confused again. Easy to do.
All known MUCs? Presumably this is all MUCs that Gajim has ever joined, or is it all MUCs it's ever discovered via disco, MUC search services, etc?
Bookmarks are a way in XMPP for clients to hold a list of group chats it knows and wants to remember. What "knowing" means, is up to the clients. Usually this means, a group chat the user was a part of at least once.
Right, I think bookmarks are a way for *users* to hold a list of chats they want to remember. My client shows them all, joined or not, in the chat roster. So to handle that, I'd now need to bookmark bookmarks...
Such a concept i would assume is als present on other chat networks.
If you login to a gateway for Matrix, you expect that your "Known Group Chat List" now also knows all the known Matrix Chats from your Matrix account.
Ah, yes, and I agree that a client should be able to discover them easily. But I think that ramming them all into bookmarks is not the right way. What you want is either: * Your client gets told, somehow, about the place to disco for more rooms. * Your client gets told, somehow, about the place to issue a XEP-055 search for more rooms. * Your server gets told that, for you, the XEP-0055 room search should also search these other places. Of these, I think I like the last one best.
The most obvious way for a gateway to sync that info would be to add them to your bookmarks. (for contacts we have the roster where the same mechanism exists)
Now you have a bunch of matrix rooms in your bookmark list, but no name for them, and no attribute for the server to put it.
I think if the data does not fit, that suggests it's the wrong data. Dave.
bookmarks are a way for*users* to hold a list of chats Certainly, but there are valid use-cases for entities to populate lists of chats without user interaction. Aren't server extensions pre-populating rosters of newly registered users pretty popular? My client shows them all, joined or not, in the chat roster. What human-readable name do you display in the UI of your client for not-joined rooms? So to handle that, I'd now need to bookmark bookmarks...
There was a misunderstanding we sorted out OOB. This protocol would let you display a friendlier name than the JID localpart of not-disco-queried(-yet?) rooms in your UI, not propose a mechanism to discover "new" rooms.
* Your client gets told, somehow, about the place to disco for more rooms.
You can also disco#items the MUC/gateway component and you do get named rooms in there, this is true. Maybe this is cleaner than using bookmarks, but it lacks a "push" mechanism. As you said, it's a matter of tradeoffs. It is likely more work for clients to nicely integrate the disco#items replies of MUC-services-that-are-gateways than to process data from a bookmark extension.
* Your client gets told, somehow, about the place to issue a XEP-055 search for more rooms. * Your server gets told that, for you, the XEP-0055 room search should also search these other places. I think room search serves a different purpose. For instance, in the case of a matrix gateway, it should map to the Matrix global room search (if there is such a thing). Here we are talking about a much more limited set of rooms, that clients "should know about" without sending additional IQs, and can do cool things with, eg fuzzy search. I don't propose this as a general way to discover new rooms. I think if the data does not fit, that suggests it's the wrong data.
Can you elaborate on why it does not fit? It's a list of chats, and sometimes the only info we have about them is a not-human-readable JID unless the client does a disco#info. Maybe clients should do these disco#info queries and avoid displaying JID localparts. In practice they don't, but maybe this is what needs to be addressed? -- nicoco
On Mon, 26 Oct 2026 at 17:51, Nicolas Cedilnik <nicoco@nicoco.fr> wrote:
bookmarks are a way for*users* to hold a list of chats Certainly, but there are valid use-cases for entities to populate lists of chats without user interaction. Aren't server extensions pre-populating rosters of newly registered users pretty popular? My client shows them all, joined or not, in the chat roster. What human-readable name do you display in the UI of your client for not-joined rooms? So to handle that, I'd now need to bookmark bookmarks...
There was a misunderstanding we sorted out OOB. This protocol would let you display a friendlier name than the JID localpart of not-disco-queried(-yet?) rooms in your UI, not propose a mechanism to discover "new" rooms.
I think, overall, that the OOB discussion solves a lot of my concerns with this, so let me summarize: Use Case: Matrix gateway. User activates a Matrix gateway, and their Matrix puppet user is in several Matrix rooms. Because Matrix doesn't have a concept of bookmarked rooms, and XMPP does (and generally uses that as "which rooms have I joined"), the gateway needs to put these somewhere to expose the information to clients. It has to put them into Bookmarks, but might choose not to auto-join. So the new bookmarks are rooms that the user has an explicit relationship with, not just rooms the user might be interested in joining. This seems a completely valid use-case, no objections from me.
* Your client gets told, somehow, about the place to disco for more rooms.
You can also disco#items the MUC/gateway component and you do get named rooms in there, this is true. Maybe this is cleaner than using bookmarks, but it lacks a "push" mechanism. As you said, it's a matter of tradeoffs. It is likely more work for clients to nicely integrate the disco#items replies of MUC-services-that-are-gateways than to process data from a bookmark extension.
This is a *different* use-case, and the one I thought was being addressed. My error - though if we can get the XEP refined later to properly capture the intended use that'd be excellent.
* Your client gets told, somehow, about the place to issue a XEP-055 search for more rooms. * Your server gets told that, for you, the XEP-0055 room search should also search these other places. I think room search serves a different purpose. For instance, in the case of a matrix gateway, it should map to the Matrix global room search (if there is such a thing). Here we are talking about a much more limited set of rooms, that clients "should know about" without sending additional IQs, and can do cool things with, eg fuzzy search. I don't propose this as a general way to discover new rooms.
Right, agreed. And, also, I think it'd be a uiseful feature to have, and I might see about a spec... But please, if someone else beats me to it that'd be even better.
I think if the data does not fit, that suggests it's the wrong data.
Can you elaborate on why it does not fit? It's a list of chats, and sometimes the only info we have about them is a not-human-readable JID unless the client does a disco#info.
Maybe clients should do these disco#info queries and avoid displaying JID localparts. In practice they don't, but maybe this is what needs to be addressed?
No, I'm wrong there. Didn't understand the nature of the data. Dave.
Hi, All the motivation I've read is about clients effectively wanting access to the current room name, but without doing a disco#info query. The XEP introduction claims that doing disco#info requires a lot of round-trips, however to me it is exactly one full round-trip from client to muc server (via own server) and back. For multiple rooms, this can be fully parallelized, thus not causing a significant increase of the time needed to do this (assuming usual broadband speeds these days). The proposed solution is to put something in the bookmarks that was the name of the room at some point, for clients to display that when no user-selected name is available. How and when the information in the bookmark is updated remains largely unspecified, presumably because it might be updated not only by clients but also by the server or a gateway component (and those potentially even ensure this is in fact the current room name and not something old). I also noticed that for this feature, the room name has been picked as a specific item from the disco#info to be handled this way, other information that is potentially relevant for a room search (specifically mentioned in the XEP) or listing (like room features, subject/description, language, number of occupants and room avatar - I just randomly picked what s.j.n displays) are not included. This means this functionality is only helpful to some clients, others will remain unsatisfied and still need to do disco#info the same as before. To me, this handpicks some specific issue we see across the ecosystem caused by the current disco approach. Maybe it makes more sense to review and improve this in a more holistic way? Marvin On Sun, 2026-10-25 at 05:28 +0000, Daniel Gultsch wrote:
The XMPP Extensions Editor has received a proposal for a new XEP.
Title: Bookmark last known name Abstract: This document defines an XMPP protocol extension to allow users to pin PEP Native Bookmarks.
URL: https://xmpp.org/extensions/inbox/bookmark-last-known-name.html
The Council will decide in the next two weeks whether to accept this proposal as an official XEP. _______________________________________________ Standards mailing list -- standards@xmpp.org To unsubscribe send an email to standards-leave@xmpp.org
participants (6)
-
Daniel Gultsch -
Dave Cridland -
Marvin W. -
Nicolas Cedilnik -
Philipp Hörist -
snit