[MPlayer-dev-eng] Re: [PATCH] Donald Graft's kerndeint adaptive deinterlacing filter

[D Richard Felker III]

On Mon, Feb 02, 2004 at 10:42:09AM +0100, Alban Bedel wrote:
> > > At least some filters do indeed print a help message and exit when
> > > called this way.
> > 
> > Yes, and Albeu drank a whole ocean of cola for that crap. It's total
> > nonsense for filters to be able to terminate the calling program.
> 
> And it's total nonsense to say that. It's just plain FALSE: RTFS or
> shut your mouth !!!!

Would you care to explain what's false? I guess the filter itself
doesn't exit mplayer, but the config-setting function does...

> > His
> > system also violates the privacy of the _priv_ structure of filters,
> > which was named _priv_ for a reason!!
> 
> yeah this suck. The reason ? A good bunch of the filters will need
> nearly identical option struct and priv struct. It was also needed to
> have 100% backward comp. for filter wich still use a simple string as
> argument.

It was not needed. All you had to do was modify the open functions to
copy the fields from a config struct into their priv struct/

> Don't be stupid there are always reasons behind these kind
> of things.

This is not a legitimate reason, and in fact there is NO legitimate
reason, EVER, for violating a priv element. In fact it might not even
be a pointer, but some other type cast to void *!

> Now if you have a better idea feel free to code it ;)

It's already been done for G2.

Rich