I just had an interesting conversation with "yagiza" about XEP-0184,
which he has said I can paste here. The general idea is: do we need
something in XEP-0184 to indicate that a message has been read by the
intended recipient? This would be similar to the <displayed/> element in
XEP-0022. I'm not convinced that we need this feature in XEP-0184, and
tried to explain why to yagiza. Further discussion is welcome on this
list. I've attached the log of my chat with yagiza for context.

/psa


*** 2010-06-16
[11:26:30] *** yagiza is Online
[11:26:23] <yagiza> Hello!
[11:27:33] <stpeter> hi, can I help you?
[11:29:18] <yagiza> Yes. Have you read tomorrow discussion on 
[email protected] about XEP-0184?
[11:29:27] <yagiza> yesterday
[11:29:33] <yagiza> discussion
[11:29:48] <stpeter> I haven't
[11:29:56] <yagiza> Ok
[11:30:21] <stpeter> I'm sure I was "in" the room, but not paying attention
[11:30:59] <yagiza> It was a dispute about meaning of <received/> element
[11:32:48] <yagiza> Some of participants told that it means that client 
received and processed message, others - that client assure that user read the 
message.
[11:33:26] <yagiza> Bombus, for example, works 1st way
[11:33:50] <stpeter> it does not mean that the user read the message
[11:33:58] <yagiza> Ok
[11:34:31] <yagiza> So, I suggested to add <read /> element to the XEP for this 
case.
[11:35:00] <stpeter> how does the client really know that the user read the 
message?
[11:35:07] <yagiza> Well...
[11:35:13] <stpeter> "Click here to verify that you have read the message"?
[11:35:57] <yagiza> Most of the clients just after they received the message, 
mark them as "new"
[11:36:26] <stpeter> the client might know that the message has been 
*displayed* but that doesn't guarantee that the message has been *read*
[11:37:04] <yagiza> Then, after they assure that user read it (for example, the 
chat window, whith the message got the focus) client marks those messages as 
"read".
[11:37:42] <yagiza> So, the fact that "new" message became "read" could be 
flagged by the <read /> message.
[11:39:23] <stpeter> but displayed != read :)
[11:39:35] <stpeter> then someone is going to ask for <displayed/>
[11:39:45] <yagiza> (^_^)
[11:39:49] <yagiza> Yes
[11:40:20] <stpeter> see http://xmpp.org/extensions/xep-0022.html
[11:40:23] <stpeter> that has <displayed/>
[11:40:45] <yagiza> Yes, but that XEP is obsolete
[11:40:52] <stpeter> so XEP-0022 differentiated between delivered and displayed
[11:41:01] <stpeter> for XEP-0184 we decided that we didn't need displayed
[11:41:07] <yagiza> Yes
[11:41:14] <yagiza> So, it should be discussed, which of them are really 
important.
[11:41:27] <yagiza> I don't think displayed is really necessary
[11:41:42] <yagiza> But I think that "read" is.
[11:41:44] <stpeter> I don't either
[11:42:00] <stpeter> and what is the UI for determining if the message has been 
read?
[11:42:16] <yagiza> As I wrote before
[11:42:27] <stpeter> that's not enough
[11:42:41] <stpeter> just because the chat window has focus doesn't mean the 
message has been read
[11:42:54] <stpeter> and just because the chat window does not have focus 
doesn't mean the message has not been read :)
[11:43:23] <stpeter> I have lots of windows open all the time and I read 
messages in inactive / unfocused windows
[11:43:47] <yagiza> Yes, but for me the fact that I don't have new messages is 
enough to think that I've read all the messages.
[11:44:03] <yagiza> So, the other side could be satisfied with that.
[11:45:32] <yagiza> And I'm always clicking on the windows which marked as 
"having new messages" just to be sure that no new messages arrived.
[11:45:45] <stpeter> so for you, <read/> *really* means "displayed in an 
interface that has focus"
[11:46:29] <yagiza> No. It is implementation dependent.
[11:46:33] <yagiza> But
[11:47:03] <yagiza> It just means, that implementation marked that message as 
"read", not "new"
[11:47:15] <stpeter> e.g. I have 15 chat tabs open in Psi right now -- I could 
quickly cycle through them all (8 of them have "new" messages, and some of 
those have 500+ messages) so that all of the messages are marked as read, but 
that doesn't mean I have actually looked at all the words and absorbed the 
meaning 
[11:47:38] <yagiza> Yeah
[11:48:18] <stpeter> so you want to know if my client has marked those messages 
as read, even if I haven't read them -- correct?
[11:48:37] <yagiza> But in bombus, for example messages marked as "read" 
different way than in Psi
[11:48:46] <yagiza> Yes
[11:48:52] <yagiza> Exactly
[11:48:57] <stpeter> heh, ok
[11:49:05] <stpeter> at least now we have a common understanding
[11:49:58] <stpeter> and my client might even have a special button that says 
"mark all messages as read" (I know my email client has that)
[11:50:07] <yagiza> Yes
[11:50:35] <stpeter> um, so what is the point of communicating to you the fact 
that I've marked a message as read?
[11:50:39] <stpeter> what do you gain?
[11:50:50] <yagiza> In the client I'm developing right now, you have to focus 
the message with cursor to mark it as "read".
[11:50:51] <stpeter> you gain a false sense that I've read the message, but 
perhaps I haven't
[11:51:23] <yagiza> In the most cases that's enough.
[11:52:25] <stpeter> ok, so... I am extremely busy this week and I don't have 
time to discuss this in depth, but you could propose a change to XEP-0184 on 
the [email protected] list or we can chat about it some more in the jdev room
[11:52:47] <yagiza> Ok. Thank you!
[11:53:00] <stpeter> it's fine with me if you copy and paste the log of our 
conversation, too
[11:53:17] <yagiza> Right now?
[11:53:34] <stpeter> e.g., if you want to attach this to an email message that 
you send to the standards@ list
[11:53:57] <yagiza> Ok

Attachment: smime.p7s
Description: S/MIME Cryptographic Signature

Reply via email to