Abstract: | This document discusses the use of XMPP Extensions to provide a useful user experience when using multiple clients. |
Author: | Kevin Smith |
Copyright: | © 1999 - 2010 XMPP Standards Foundation. SEE LEGAL NOTICES. |
Status: | ProtoXEP |
Type: | Informational |
Version: | 0.1 |
Last Updated: | 2010-08-12 |
WARNING: This document has not yet been accepted for consideration or approved in any official manner by the XMPP Standards Foundation, and this document is not yet an XMPP Extension Protocol (XEP). If this document is accepted as a XEP by the XMPP Council, it will be published at <http://xmpp.org/extensions/> and announced on the <standards@xmpp.org> mailing list.
1. Introduction
1.1. Behaviour
2. Typical Use
3. Specialized Use
3.1. Stateless clients
3.2. Bandwidth-Constrained Clients
Appendices
A: Document Information
B: Author Information
C: Legal Notices
D: Relation to XMPP
E: Discussion Venue
F: Requirements Conformance
G: Notes
H: Revision History
Historically, there have been many problems with the experience when users have used multiple clients - either because they have multiple clients connected at once (e.g. a client on their phone and a client on their workstation), or because they use different machines at different times (e.g. a laptop when away and a desktop when at work). XEPs exist that can address these issues, and this Informational [proto]XEP documents a sensible approach to their collective use for providing a consistent user experience on all their devices or clients.
For examples of benefits, if their clients follow the recommendations in this document, a user will:
You would use message archiving to have a local message cache of the history (This is going to be needed for reasonable search and stuff, and consistent with things like Thunderbird's approach to IMAP), and that's just a case of "Server, send me all messages since timestamp X (my last archived message in the cache)" and it sending forwarding encapsulated messages and then saying Done. The client enables carbons. The client asks all other resources for any unread messages. As an unread message comes in, either from a carbon or directly, the client can then create a new chat window, and populate it with context from the local history cache. If an outgoing carbon comes in from a given resource, the client can mark as read all messages from the same jid to that resource, to stop having different clients look like they've got unread messages, when they were handled elsewhere.
Some clients will not be able to implement all the recommendations due to operating constraints. In such cases it's recommended that they get as close to the Typical Use recommendations as possible. Suggested compromises for some cases are given below.
If a client has limited (or no) storage available, it may be unable to persist a complete history of the user's conversations. In this case, it's recommended that the client still requests history from the message archive, but that instead of requesting a full history (asking for all messages that it hasn't yet seen), it requests a small amount of history (e.g. a week for a client with little storage, or an hour for a client with no storage) and stores this where it can (either in storage or in memory), and that it purges history older than this (either from storage or from memory, accordingly).
Some clients will either wish to, or be able to, only use a little bandwidth, or to use it infrequently (e.g. those running on mobile phones). In these cases, it's recommended to request limited history (as for Stateless Clients), and also (if necessary) to not enable Carbons while they know the user is not actively attending to the client. This means that a client running in the background on a phone may turn off Carbons to avoid frequent bandwidth use while the user potentially chats on another device. When the client then comes to foreground, or gains the user's attention, the client may then enable Carbons, and also request unread messages from other clients. This means that the user may not be alerted to incoming messages directed at another client until the constrained client is in the foreground, so this should only be enabled if it's deemed necessary.
Series: XEP
Number: xxxx
Publisher: XMPP Standards Foundation
Status:
ProtoXEP
Type:
Informational
Version: 0.1
Last Updated: 2010-08-12
Approving Body: XMPP Council
Dependencies: XMPP Core
Supersedes: None
Superseded By: None
Short Name: multiple-clients
This document in other formats:
XML
PDF
Email:
kevin@kismith.co.uk
JabberID:
kevin@doomsong.co.uk
The Extensible Messaging and Presence Protocol (XMPP) is defined in the XMPP Core (RFC 3920) and XMPP IM (RFC 3921) specifications contributed by the XMPP Standards Foundation to the Internet Standards Process, which is managed by the Internet Engineering Task Force in accordance with RFC 2026. Any protocol defined in this document has been developed outside the Internet Standards Process and is to be understood as an extension to XMPP rather than as an evolution, development, or modification of XMPP itself.
The primary venue for discussion of XMPP Extension Protocols is the <standards@xmpp.org> discussion list.
Discussion on other xmpp.org discussion lists might also be appropriate; see <http://xmpp.org/about/discuss.shtml> for a complete list.
Errata can be sent to <editor@xmpp.org>.
The following requirements keywords as used in this document are to be interpreted as described in RFC 2119: "MUST", "SHALL", "REQUIRED"; "MUST NOT", "SHALL NOT"; "SHOULD", "RECOMMENDED"; "SHOULD NOT", "NOT RECOMMENDED"; "MAY", "OPTIONAL".
Note: Older versions of this specification might be available at http://xmpp.org/extensions/attic/
Initial version.
(ks)END