[Standards] Update to MIX 0.9.7

Steve Kille steve.kille at isode.com
Wed May 9 13:53:43 UTC 2018


I have just issued a PR for MIX 0.9.7

This covers a number of things

First, the recent discussion initiated 0n 23 April by Manuel Rubio on room
creation, with input from Ralph Meijer and Jonas Wielicki.   I have resolved
all this as indicated in the email thread.

In reviewing this change,  I noted that most configuration attributes have
sensible defaults, but "nodes present" did not.  I have added a sensible
default.   I have also clarified handling of "last change made by".   

Florian Schmaus sent a message with a large number of points on 20th March.
Dave Cridland and Kevin Smith sent detailed responses.

This update addresses the detailed comments in his message, mostly in line
with his suggestions and subsequent comment.    Notes below where I have not
made changes and information on non-obvious changes.

1.   I have retained use of type=groupchat.  Dave Cridland made a clear
argument to retain this, which Kev and I agree with.   I've not seen any
other views stated.

2.   Requiring MAM archive on the user's server.   I am not changing the
current text, which makes this mandatory.   Kev notes that this will be
addressed in XMPP2.   Making this optional would make things trickier for
the client (which could no longer rely on MAM archive in the local server
for joined channels).

3.  Mandatory presence (3.9.7).   There is an option for a MIX  channel to
require presence.  This allows a channel to specify the current MUC
behaviour that online clients are visible with presence (and no "hidden"
listeners, which some might object to on privacy grounds).   This cannot be
enforced by the MIX channel, so it is a policy that compliant MIX clients
are expected to follow.   I have clarified this in the text.   It seems
useful to me, but we could drop this option if people feel it will never be
useful.

4. Florian Schmaus  noted on 6.1.2 about annotation.   I've made some minor
edits to 6.1.2, but key change on this made to 7.8.   Annotation only
affects client/local server communication.

5.  6.3 (Ensuring Message Delivery) describes an important function for MIX.
The detailed approach has issues, which Florian Schmaus flags.   Jonas
Wielicki also flagged the issues in Feb 2017.   I am replacing this section
with a reference to a (yet to be written) XEP.    Rationale:
    - We clearly do not have the spec right
    - Reliable message delivery seems like a generic capability that could
be used elsewhere.


6.   6.5.1.   Kevin Smith suggested that no change was needed, so I have not
made any change.




I've also updated all MAM URNs to urn:xmpp:mam:2  (thanks to Manuel Rubio
for pointing this out)


Florian made some broader comments.

I do not agree with the assertion that MIX is bloated.   It provides a
similar function to MUC and is of comparable size to the MUC and PubSub
specs.    

It would be nice to split MIX, but there are no major chunks that I can see
viable to separate.   I do not see any way to achieve a "slim core".
There are a few small pieces that could be shaved off the edges into
separate XEPs, but this is a mass of editorial work and I don't think we
should so this unless there is clear reason.    I have move 6.3 (Ensuring
Message Delivery) out, and am happy to review other specific suggestions.


Florian flagged two "controversial points":
   - Channels in Roster
   - JID Hidden Channels

I have added new sections, which I hope will add clarity on these points and
why the choices were made.


I think there remains a broader issue that people come to MIX expecting it
to be like MUC and the differences, such as the two above, lead to
confusion.    I think that this has also contribute to a view that MIX is
complex (it really is not that complex), because it is different to MUC.  To
address this,  I've added a section which introduces MIX to those familiar
with MUC.   I believe this will be helpful now, although we may choose to
drop in the future.



Steve





More information about the Standards mailing list