Re: [gtk-vnc-devel] Regression tests

["Daniel P. Berrange" <berrange redhat com>]

On Tue, Jan 15, 2008 at 02:10:16PM +0000, Richard W.M. Jones wrote:
> Daniel P. Berrange wrote:
> >On Tue, Jan 15, 2008 at 01:49:55PM +0000, Richard W.M. Jones wrote:
> >>Have you ever discussed having regression tests for Gtk-VNC?  I've been 
> >>able to provoke at least one segfault with a test program (program + 
> >>patch coming up).  I wonder if we should keep such test programs around 
> >>to test for regressions?
> >
> >We've not really considered tests before - any ideas / suggestions welcome.
> >By test program, do you mean a fake VNC server to run against, or were you
> >testing & crashing client side APIs ?
> 
> Yes, just after posting this I realised we'd need a fake/test standalone 
> VNC server.  Looking around on the 'net, such things don't seem to 
> exist.  Perhaps we could wire up Xvfb somehow.

There's probably a couple of different scenarios we'd need to try and
test. Using Xvfb (or just Xvnc) and simulating use of various apps to
validate GTK-VNC against a real VNC server impl. The other scenario
would be to write a test server which speaks RFB protocol and just 
sends back various pre-programmed sequences of RFB messages. We could
write some of the test sequences to send delibrately illegal RFB 
messages to test our malicious server handling, eg a hextile update
for a 16x16 region at x=900,y=900, on a desktop only 800x600 in size.
Or send back 100000 auth types. Or a negative sized update. Etc, etc
I know a regular VNCviewer client crashes & burns on some of these
becasue QEMU used to have bugs where it'd send out-of-bound updates
and crash the viewer :-)

Dan.
-- 
|=- Red Hat, Engineering, Emerging Technologies, Boston.  +1 978 392 2496 -=|
|=-           Perl modules: http://search.cpan.org/~danberr/              -=|
|=-               Projects: http://freshmeat.net/~danielpb/               -=|
|=-  GnuPG: 7D3B9505   F3C9 553F A1DA 4AC2 5648 23C1 B3DF F742 7D3B 9505  -=|