Oleg:
Heya. A quick bug report:
1. I compiled the 12 Feb 02 deliverable on a WinNT machine, and
when I start it up with another Kaboodle instance already
running on the LAN, it crashes. Debugger takes me to
NIDDevice.cpp::Serialize(), the line right under the IKO
comment: "Potential Bug". I guess we should upgrade that to
"Confirmed Bug". :) Only interesting thing in the locals
list is numChildren: 1699749996. Lotsa children.
2. This time, I startup Kaboodle on the WinNT machine in debug
mode, with no other Kaboodles on the LAN. After it discovers
the LAN, I boot another WinNT on the LAN. As soon as I login
on that machine, Kaboodle crashes. This time the debugger
takes me to NIDNetwork::GetNetworkGwyID(), where (as we've
seen before), the 'this' pointer is having trouble. It's
value is 0xdddddddd.
Interestingly...the other WinNT machine had an old
copy of EchoFree on it (the one from the ftp server). It wasn't
active, but apparently was interfering with Kaboodle. When I
de-installed it, Kaboodle didn't crash immediately. I had to
startup Kaboodle on *that* machine to crash in the same place
(in NIDNetwork::GetNetworkGwyID()).
It strikes me that Kaboodle is *very* sensitive to the
formating of the UDP broadcasts it uses to communicate to other
instances on the LAN (those UDP heartbeats it sends out to
255.255.255.255 port 4082 every 2 seconds). Shouldn't there
be a "validity" checker on that socket read, to verify it's
well-formatted Kaboodle data and not just something random?
-Scott
On Tue, 12 Feb 2002, Lyapin Oleg wrote:
> Section13 of Spec is done (Unify draw functionality of Kaboodle dialog
> panels).
>
_______________________________________________
Kaboodle-devel mailing list
[EMAIL PROTECTED]
https://lists.sourceforge.net/lists/listinfo/kaboodle-devel
