[spook-l] Features and bugfix request

[Nathan Lutchansky]

On Mon, 2005-11-07 at 08:28, Vladimir Lasky wrote:
> G'day Nathan,

Hey Vlad...

> I have some questions regarding features and bugfixes:

I haven't spent any time on Spook for a while; all my effort has been
going into a new media library that the next version of Spook will be
based on.

> 1. How hard would it be to implement an option to allow the image to be rotated
> 90 degrees clockwise/counterclockwise, i.e. so you could transmit a
> portrait view rather than a landscape view.

Not hard, it's just a matter of transposing the image.  There's lot of
little options like that that need to be added.  Cropping, rotation,
mirroring, etc.

> 2. Some users are really being put off by the dirty-looking green inter-frames
> that are visible for the first 10 seconds when viewing a video stream using
> Quicktime. Is it possible to fix this, perhaps as simply as waiting until the
> next complete keyframe is ready before commencing streaming?

Yeah, that's kind of annoying, but I haven't found an easy way around
it.  If the stream is delayed, say, four seconds until the next
keyframe, QuickTime will increase the buffering delay from the default
of three seconds to seven seconds.  That seems more annoying than the
current behavior.  I think the appropriate workaround would be to delay
the response to the PLAY command until a keyframe is about to be sent,
but there's not really a clean way to do this with the current design.

> 3. I have noticed that spook generates a Floating point exception if you run it
> without all your configured hardware connected e.g. When spook.conf has three
> cameras configured, but only two are connected at the time spook is executed.

Is this with the IIDC input module?  The multi-cam support wasn't
particularly clean to begin with.  I'd have to test it with my own
system and that will have to wait until I'm back in my office in a few
weeks.

> 4. The IEEE 1394 for Linux people (www.linux1394.org) state that kernel 2.6.12
> to be the first stable firewire release and that many issues that generated
> warning messages in the past have now been resolved. Perhaps it's worth
> advising people to use versions from this one onwards?

I think the 1394 support in the kernel has been pretty stable for a
while now, and I haven't seen any kernel-related problems in years.  Are
there specific issues that are resolved in 2.6.12 that would impact
Spook?  -Nathan