[mythtv] osd problem could be compiler

[Isaac Richards]

On Saturday 08 November 2003 12:29 pm, thor wrote:
> On Saturday 08 November 2003 10:27, Tony Clark wrote:
> > > I just checked in some changes that seem to have fixed things for me.
> > > I'd appreciate it if other people could test the fix as well.
> >
> > Yep that seems to have fixed it.  I haven't tested with gcc-3.3.2, only
> > gcc-3.2.3 but I guess it is ok.
>
> 	Solved the problem here with gcc 3.2.2., although I have no idea why that
> "fix" should have solved anything.  =)

Just explained a little about this to Thor on the IRC channel, but figured 
others might be vaguely interested:

I'm leaning towards it being an optimization bug with respect to a class 
inheriting from an instantiated template class and having the complete class 
definition in a header file.

Basically, I got lucky -- the first thing I did when I went to start debugging 
after finding that I could reproduce it with gcc 3.2.3 was to want to put 
various debugging statements in the ProcessFrame function without having to 
recompile multiple files every time I made a change.  So, I moved the 
function into the filtermanager.cpp file from the filtermanager.h, and the 
OSD came back =)

Isaac