-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

On 8/26/09 6:03 PM, Florian Zeitz wrote:
> Peter Saint-Andre wrote:
>> Today in the xmpp:[email protected] room we had a bikeshed
>> discussion about the various transitions in Chat State Notifications
>> (XEP-0085)...
> 
>> http://logs.jabber.org/[email protected]/2009-08-26.html#15:12:36
> 
> One quick note about this: Some people have mentioned during the
> discussion they find this whole XEP pointless 

Really? I have not heard that. It is quite widely implemented.

> and calling this a
> bikeshed discussion has a similar impact. 

Quite possibly. In fact, nothing in the spec right now forbids any state
transition, so I think that the discussion is a bit of a time-waster.
However, I've made some proposed text clarifications to the spec anyway.

> I think it implies that this
> is not really worth discussing, which automatically degrades any answer
> to the question, which could probably be seen as an insult to somebody
> who does care, or even who bothers to join the discussion.
> That said I personally don't care that much either, the following is
> just to put another point of view out there.

Thanks for your feedback.

>> I suggest that the following are the most common/sensible transitions:
> 
>>                 o (start)
>>                 |
>>                 |
>> INACTIVE <--> ACTIVE <--> COMPOSING <--> PAUSED
>>     |                                       |
>>     |                                       |
>>     +---<---<---<---<---<---<---<---<---<---+
> 
>> Someone suggested that you might want to do things like go from
>> <inactive/> to <paused/> if a user returns to a chat session interface
>> containing an unfinished message. I have no deep objection to such a
>> transition, though it strikes me as a bit odd. My reasoning is that the
>> <active/>, <inactive/>, and <gone/> states refer to the overall chat
>> session interface whereas the <composing/> and <paused/> states refer to
>> the message input interface (and are in some sense a subset of
>> <active/>, so that you would go from <paused/> to <inactive/> but from
>> there back to <active/> and then <composing/>).
> 
> I would return to <paused/> for the same reason. <paused/> means someone
> has been composing  and is active (both intentionally without </>),
> which to me is obviously the case if the message input interface is not
> empty.

I must say this feels quite unnatural to me, but if an implementation
would like to proceed that way then I don't think the spec should forbid
that (which is currently doesn't, but my proposed text changes make that
even clearer).

>> As I said this is painting the bikeshed and I'd just as soon leave the
>> supported state transitions up to the implementation so that we don't
>> need to argue about the spec all the time, but if people care about this
>> I will update the spec.
> 
> As said in the discussion I think that might be the way to go if
> implementors stand by their different opinions.

Yes, which is why I made the text changes I did.

Given that no one involved in the original discussion (triggered by a
bug report against the Gajim client) has posted in this thread, I must
admit that I wonder how passionate they are about correcting the spec.
However, I think the text changes I propose address their concerns.

Peter

- --
Peter Saint-Andre
https://stpeter.im/


-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.8 (Darwin)
Comment: Using GnuPG with Mozilla - http://enigmail.mozdev.org/

iEYEARECAAYFAkqoYvcACgkQNL8k5A2w/vxoFACg8nn+wlV5FOaIWD4yicUZU0jh
5R4AoNVyKASsDCozR+4ANkLTnsGfAiQl
=Ld+u
-----END PGP SIGNATURE-----

Reply via email to