[hatari-devel] Raster stability issue in new Hatari

[npomarede at corp.free.fr]

On Sun, 12 Jul 2009, Eero Tamminen wrote:

> Hi,
>
> I noticed that in current Hatari the rasters are very unstable in this
> Automation disk menu:
> http://ataristgames.free.fr/AUTOMATION/500-5xx/A_500/A_500_B.ST
>
> (I was trying to verify one of the issues on the compatibility list.)
>
> In Hatari v1.0 there's no unstability.
>

Hello,

this problem occurs when spec512 mode is not used to change the color at 
the exact cycle.
In that case, Video_SetHBLPaletteMaskPointers determines the line where 
the palette should be modified to use this new color.

The problem is that with the new video.c, I considered that any change 
before cycle 512 applied to line nHBL, but in fact as the maximum possible 
display position is cycle 460, we should use cycle 460 as a limit to 
decide if the change applies to line nHBL or nHBL+1 (this was handled in 
previous video.c version, but the formula used LineStartCycle, which 
worked, but was not the correct limit).
Else, any change close to +/- cycle 512 was applied to line nHBL or 
nHBL+1, which gave this ugly flickering.

The "problem" is that automation menu 2/500 is in med res, and we don't 
have support for spec512 mode in med res. While this bug would not have 
shown in lo res (because hatari would switch to spec512 mode), it was 
visible in med res. Thanks for reporting this.


Anyone feeling like implementing spec512 mode in med res ? :)

Nicolas