Re: Presence issues

2009-06-01 Thread Martin Atkins


We did have some presence bugs related to this a long time ago, but I 
thought they were fixed now. You're running on trunk, or at least the 
latest stable release, right?


Gavin Mogan wrote:

How to reproduce:

1) Login to server using miranda jabber client
2) Login to server with different connection (miranda again works, i
use psi for my tests though)


Client #2 doesn't see client #1 online until client #1 sends another
presence indicator.

I've tried this with other combinations, and most of the time server
sends out a presence message on behalf of the other clients online on
your roster, but not miranda.

Miranda:
presence
priority5/priority
c xmlns=http://jabber.org/protocol/caps;
node=http://miranda-im.org/caps; ver=0.7.9.0 ext=secureim mood /
status/status
/presence

PSI:
presence
priority5/priority
c xmlns=http://jabber.org/protocol/caps;
node=http://psi-im.org/caps; ver=0.12 ext=cs ep-notify html /
/presence


I'm not really sure where to go from here, I tried to walk though but
I get lost in. I'm getting tempted to actually wrap every sub hit to
track the flow a bit to find out where things are going.
I'd appreciate some help debugging or a flow of how things work or something.




Re: Presence issues

2009-06-01 Thread Gavin Mogan
gav...@gavinm djabberd $ svn info
Path: .
URL: http://code.sixapart.com/svn/djabberd/trunk
Repository Root: http://code.sixapart.com/svn/djabberd
Repository UUID: 43dd9337-660f-0410-bd32-e7601923a1a1
Revision: 867
Node Kind: directory
Schedule: normal
Last Changed Author: mart
Last Changed Rev: 867
Last Changed Date: 2009-04-29 16:54:32 + (Wed, 29 Apr 2009)

svn status doesn't seem to indicate any related files being changed
(got one in MUC, Authen/Pam, and a couple debugging lines in
DJabberd/djabberd)
On Mon, Jun 1, 2009 at 4:48 PM, Martin Atkins m...@degeneration.co.uk wrote:

 We did have some presence bugs related to this a long time ago, but I
 thought they were fixed now. You're running on trunk, or at least the latest
 stable release, right?

 Gavin Mogan wrote:

 How to reproduce:

 1) Login to server using miranda jabber client
 2) Login to server with different connection (miranda again works, i
 use psi for my tests though)


 Client #2 doesn't see client #1 online until client #1 sends another
 presence indicator.

 I've tried this with other combinations, and most of the time server
 sends out a presence message on behalf of the other clients online on
 your roster, but not miranda.

 Miranda:
 presence
 priority5/priority
 c xmlns=http://jabber.org/protocol/caps;
 node=http://miranda-im.org/caps; ver=0.7.9.0 ext=secureim mood /
 status/status
 /presence

 PSI:
 presence
 priority5/priority
 c xmlns=http://jabber.org/protocol/caps;
 node=http://psi-im.org/caps; ver=0.12 ext=cs ep-notify html /
 /presence


 I'm not really sure where to go from here, I tried to walk though but
 I get lost in. I'm getting tempted to actually wrap every sub hit to
 track the flow a bit to find out where things are going.
 I'd appreciate some help debugging or a flow of how things work or
 something.




Re: Presence issues

2009-06-01 Thread Martin Atkins

Gavin Mogan wrote:

Ah, i think it might be the JID case sensitivity issue again, was it
ever merged?

I'm seeing that my roster sent testuser@ and i'm logged in with TestUser@



No, that was not addressed. I don't think there was consensus on how 
best to handle backward compatibility.


I'm happy to merge it if I'm mistaken and there was some consensus on this.



Re: Presence issues

2009-06-01 Thread Gavin Mogan
I think its interesting to point out that some code detects this and
others doesn't.

In my above example, client #2 eventually sees client #1 online when
the next presence tag is sent to the server.

And it works if client #2 comes on first, then client #1

My psi (client #2) and my roster building code all are in lower case.

On Mon, Jun 1, 2009 at 5:16 PM, Alex Vandiver ale...@bestpractical.com wrote:
 On Mon, 2009-06-01 at 17:07 -0700, Martin Atkins wrote:
 No, that was not addressed. I don't think there was consensus on how
 best to handle backward compatibility.

 I'm happy to merge it if I'm mistaken and there was some consensus on this.

 I've not seen anyone actually complain about backwards compatability.
 But my most recent patchset includes both conversion tools to upgrade
 those who are being silently bitten by this, as well as a flag which can
 be enabled to preserve the current broken behavior, if they so choose.
 I don't recall anyone being explicitly dissatisfied with the those
 options for dealing with the backcompat issue.
  - Alex




Re: Presence issues

2009-06-01 Thread Alex Vandiver
On Mon, 2009-06-01 at 17:20 -0700, Gavin Mogan wrote:
 I think its interesting to point out that some code detects this and
 others doesn't.
 
 In my above example, client #2 eventually sees client #1 online when
 the next presence tag is sent to the server.
 
 And it works if client #2 comes on first, then client #1
 
 My psi (client #2) and my roster building code all are in lower case.

If clients apply nodeprep themselves, then they'll figure it out.  If
they don't, they'll just silently drop the stanza.  The RFC says the
SHOULD apply nodeprep, (while the server MUST) which means they're free
to assume it's already kosher if they choose. :/
 - Alex



Re: Presence issues

2009-06-01 Thread Peter Saint-Andre
-BEGIN PGP SIGNED MESSAGE-
Hash: SHA1

On 6/1/09 9:33 PM, Gavin Mogan wrote:
 Does that really apply to presence. The client just sends a
 presence/presence packet to the server. The server decides who to
 send it to.
 Doesn't it?

Correct.

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

iEYEARECAAYFAkokoLoACgkQNL8k5A2w/vySkgCgtEArfYfZ4fbguD0YRkTGK+uL
pnMAoI4AQ0F+Dns9vyxZlN8Vlkjgvmxf
=xIbu
-END PGP SIGNATURE-