Re: [Gnash] gnash testing

[Rob Savoye]

Jens Petersen wrote:

> For me the client from current anoncvs head runs ok under FC4 but not on 
> FC5: I noticed Fedora Extras 4 has gtkglext 1.0 whereas Extras 5 has 
> 1.2, but I don't know if that is the cause of the crash I see under FC5 
> whenever I run gnash on any flash file:

  My Fedora machine is at core 4, and I've heard enough problems with 
core 5 that I'm going to wait a bit before upgrading... If you guys at 
Redhat happen to have any FC5 machines of any architecture I can ssh to, 
it'd help with testing...

> I didn't manage to get a backtrace with gdb, perhaps it is crashing 
> before there is a stack.

  Are you running a debug build of Firefox, and then starting it with 
-g ". With Gnash, you can also invoke the Flash movie and add 
"?waitforgdb=yes", and the plugin will block in the very being of it's 
initialization. Then you can attach GDB, and set breakpoints in that 
instance.

  On FC4, I've had problems with ATK, of all things. GTK+ has been a 
bit of a moving target lately, so my gut feeling is this would be an 
issue with one of the dependent libraries.

> However on FC4 I noticed that gnash tends to grab the CPU for me even 
> after a flash file has apparently finished running.  Admittedly I don't 
> have accelerated 3D graphics, but it looks like a bug of some kind to me.

  The polling loop continues after playing to handle events. This will 
go away after the alpha release, cause I'm going to rewrite the event 
handling to be more portable and handle conditions it was never designed 
for.

> I haven't managed to get the plugin working properly in firefox.
> Currently in FC4 it just seem to display the initial flash file 
> background, but then nothing happens.  (In FC5 the plugin crashes 
> firefox as soon as it tries to render some flash.)  I haven't tried i386.

   Hum, on my i386 FC4 laptop, it works just fine.

> I attach the rpm .spec file and a configure hack I've been using to 
> package and test it.

  I'm not a fan of using pkg-config, as it doesn't exist on any 
non-GNU/Linux system, and many people don't even have this installed it 
turns out. So at one point I rewrote the configure tests for Gnash to do 
my own searching and not use pkg-config. So if you are having a 
configure problem, I'd prefer to fix my test, and now add a dependency 
for pkg-config.

  I merged in a few of your spec file changes with the other patch I 
got for the spec file. I'll check this after I do a little RPM building 
testing.

        - rob -