Oleg:
Hello! I compiled the 0.77 you sent. I like
the updates you made to the VNC Property Tab!
Two quick bug reports:
1. The VNC settings are still not "sticking" on Windows98.
I can *run* the VNC viewer thru Kaboodle from Win98 to
a WinNT machine, but the settings I make (such as the
path to the viewer .exe) don't survive a restart.
2. The VNC halo appears at odd times. It's there with the
old EchoFree halo when I first startup. Then, as soon as
I click a dependant device (one that used to give a check
mark) and deselect it by clicking in the field, it goes
away and the standard Kaboodle halo appears. But when I
click on a managed device (such as a printer) what used
to give me a checkmark now gives the VNC halo, along with
the old EF halo.
Also, I took out "PreAlpha" from the SaveNID()
class in NIDManager. I can't quite figure why it writes
the log files to "../." relative to where the Kaboodle
binary is, rather than to "./" as you'd expect. Shrug.
Anyhow, thanks for any attention you can give to
the two bugs.
-Scott
On Tue, 29 Jan 2002, Lyapin Oleg wrote:
> 29 Jan 2002 Status report
> --------------------------
>
> 1. VERSION CONTROL
>
> >I'd be happy to re-setup the CVS server, if you'd
> >prefer that? I could install the 0.77 snapshot I just sent.
>
> I am afraid we are not ready for that kind of suggestion so
> far. It is important for us to keep the intermediate file versions.
> Besides, the corporate project policy require that the
> projects are stored at a predetermined location.
>
> 2. BUGS
>
> 2.1
> > Quick bug report. Am testing the VNC tunneling
> > capabilities from a Windows98 machine to a
> > WindowsNT machine, and Kaboodle is not working
> > correctly on the Win98 machine.It's not letting me startup
> > a VNC viewer session to the VNC server
> > ...
>
> Fixed.
>
> 2.2
> >> We found 6 places from where CNIDPC::GetPCPriority
> >> is called. And, what it interesting, this function has a primitive
> >> body: { return this->m_dwPCPriority; } (it just returns a
> >> member value) so the only reason of crash can be invalid
> >> "this" pointer. Actually, the address you gave (0xC0000005)
> >> is rather high and most likely is garbage.
> >
> >Right, "this" has a null value when the program
> >crashes, according to the debugger. So I bet you're right
> >with the initialization comment.
>
> "this" hardly ever has a null value. I guess it rather has a
> 0xC0000000 value, but it does not matter anyhow.
> After a more close consideration, this problem seemed to be
> a binary version incompatibility problem. Most likely, it occurs
> when the version 0.72 tries to read socket data meant for version
> 0.75 or higher. In version 0.72, no socket data validation is
> performed before reading and so the consequences are unpredictable.
> What we see to be done in order to overcome the difficulties is to
> set a version trace mechanism. It may work like that: a special
> version key is stored in the registry and is carried over as the net
> auto-detection is performed along with other machine-related info.
> Thus, a remote machine gets to know what it may expect from that
> partner machine. The cases of version incompatibility are detected
> and no attempts are made to interpret unknown data, thus avoiding
> crash. (Of course, the old version [0.72] Kaboodle machines will
> cash anyway when working together with the new version machines).
> Let me know if you found this improvement useful and then we'll
> include it into the work plan.
>
> -Oleg
>
>
>
>
_______________________________________________
Kaboodle-devel mailing list
[EMAIL PROTECTED]
https://lists.sourceforge.net/lists/listinfo/kaboodle-devel
