[MPlayer-dev-eng] new changes to input trash terminal on exit

[Alban Bedel]

Hi D Richard Felker III,

on Sun, 24 Feb 2002 10:53:54 -0500 you wrote:

> On Sun, Feb 24, 2002 at 12:42:11PM +0100, Alban Bedel wrote:
> > Hi D Richard Felker III,
> > 
> > on Sun, 24 Feb 2002 00:03:33 -0500 you wrote:
> > 
> > > On Sat, Feb 23, 2002 at 11:02:44PM -0500, D Richard Felker III wrote:
> > > > some of the new changes in the input system today broke mplayer so it
> > > > trashes the controlling terminal on exit (it leaves echo mode
> > > > disabled when it exits). this should be fairly simple for alban to fix
> > > > since he's familiar with the changes he made, but i'll take a look and
> > > > see if i can fix it if he's not around now.
> > > 
> > > ok, it appears to be fixed now, but i had to disable an early
> > > initialization of getch2 that was added in alban's commit. i'm not
> > > quite sure why this was there -- it seems to work without it. if this
> > > needs to be put back, then more code needs to be put in place to
> > > prevent multiple initializations of getch2 (which overwrite the saved
> > > terminal status that needs to be restored on exit), or else mplayer
> > > needs to be changed so that getch2 is only initialized once at
> > > startup, and then disabled only once at exit. imo this latter solution
> > > makes a lot more sense, but i didn't want to make large scale changes
> > > like that to fix this simple bug.
> > Sorry, sorry. I putted getch2 init there because it's needed to be able to interrupt
> > network connection or cache filling. I forgot to remove the other when I merged
> > my version with the CVS one :(
> > It was the other that had to be removed, but you couldn't know it.
> 
> well getch2_disable is also called at the end of each movie in the
> playlist, so you need to get rid of that too if you want to only
> enable getch2 once at startup, and then put getch2_disable somewhere
> where it will always be called when mplayer exits. also, you might
> want to check how all this code affects mencoder, which doesn't use
> the input subsystem but of course does use the demuxers.
> 
I wanted to be able to interrupt network connection and cache filling (very
useful when you set a big cache and the network is way too slow), thats why
input met mpdemux. Anyway it shouldn't mess up mencoder, and libmpdemux
is still independant from input (ie can compile it without).
	Albeu