[standards-jig] Version 0.6 of JEP-0045 (Multi-User Chat)

Pasi Pärnänen pasi at parnanen.net
Mon Sep 23 10:27:21 UTC 2002


Nick, 
I hope I can put your mind at ease with this posting.


In reality there is always some kind of interaction between the
standardizers and the implementors. This has been very noticable in the
creation of HTML. In that case the work of creating extensions to the
protocol was left to the implementors and both Microsoft and Netscape
took all the possible advantages of that situation. If there is a need
for extensions then they will be added, whether a standard exists or
not. JEP-45 is in itself a result of implementations demanding changes,
isn't it?

Although we disagree on other matters i hope we can agree on one thing;
that the work of deciding matters regarding a protocol should be done by
the standards body, not by the implementors. We all suffer from he
mishmash of a "standard" that Microsoft and Netscape created for us in
HTML. Let us be more like the w3c from the start with muc2.

We could indeed emulate a single interaction for a treeview of available
rooms with a spidery climb around rooms, if we are strict. But wouldn't
a single request be more elegant?

At the minimum there would have to be a standardized notation for
addressing a specific room in a hierarchy relatively as well as
absolutely.

Nobody would be forced to use a hierarchy but the possibility of
implementing one would not be eliminated either.


One other thing...

We've been experimenting with machine translations and it would be great
if there could be a method to retrieve settings regarding the
environment of a room, such as default language.

Or is that considered a room feature? 


-- BlackDog
Better a sharp adversary than a dull friend.



Nick:
> Who says that we need a hierachy? What about a hierachy is so important 
> when it comes to this protocol? The protocol should _not_ reflect the 
> implementation, yet that is exactly what you are advocating. By making 
> the protocol implementation specific you are basically killing any other 
> ideas or improvements to the protocol. What if the implementation 
> suddenly has a design flaw? Would you suggest going back and altering 
> the protocol to meet application specs? If, so I don't you want you 
> anywhere near any projects I'm working on.
> 
> BlackDog:
> > The protocol has to handle parent/child relationships between rooms.
> > That's it basically. Without that there's no chance to implement any
> > hierarchy.





More information about the Standards mailing list