[vdr] Re: corrupted OSD colors (was: Setting the OSD colors; method cPalette::Colors)

[Peter Dittmann <peter.dittmann@freenet.de>]

On Fri, 01 Feb 2002, you wrote:
> On Thursday 31 January 2002 21:51, you wrote:
> > Anyway, your mentioned part of the code is suspective to me too:
> 
> Hi Peter,
> 
> sorry, but as I got no answer to my message I thought nobody was interested. 
> Later I was lucky, because the problem I found is not a problem. It was a 
> misunderstanding of myself, which came because the method looks so 
> complicated for the simple thing what it does.
> In fact the code is absolutely right, because not only one palette entry is 
> transmitted with every call, but multiple entries with one call and in case 
> of VDR all entries with one call.
> 
> A big sorry for waisting your time,
> 
> Mike

Hi Mike,

It was no waste, at least now I know now two new things that doesn't solve the
problem.
Anyway it does not answer why they mark the color after the transmitted
palette block also as fetched:

in dvbosd.c (near line 63):
const eDvbColor *cPalette::Colors(int &FirstColor, int &LastColor)
{
.......

Shouldn't it be:
 fetched[LastColor - 1] = true;  // instead of fetched[LastColor] = true;
<if I'm right LastColor is already incremented in the loop as the remark
 in code says>
This seems to mark all transmitted colors in the palette as being fetched
allready?
=> this doesn't solve the OSD problem, but what should the code realy do?


OSD color problem:
I don't know if I may repeat some earlier postings, but this is what I've
observed regarding OSD color problems:
- will happen in conjunction to high latency reacting on LIRC (serial)
   RC commands.
- will allways happening when OSD is opend; moving in the menues will
   never cause it
- problem persists on restarting VDR, driver must be restarted to cure problem
- in most cases the colors will have a high factor of transparency, 
   mostly base colors seems to preserved or just a bit shifted
- when the problem is there and the OSD is called, the colors seems to flash
   for a short duration (x ms) with the colors it should have, then the colors
   fade out

May be this helps somebody to isolate some other suspective code (I'm now
convinced we are talking about a driver behaviour not about VDR)


regards	Peter