Path: utzoo!attcan!uunet!samsung!uakari.primate.wisc.edu!ames!zodiac!jdevries
From: jdevries@zodiac.ADS.COM (Jeff De Vries)
Newsgroups: comp.sys.mac.programmer,ads.mac
Subject: PaletteManager vs. CopyBits (long; gurus only)
Message-ID: <10713@zodiac.ADS.COM>
Date: 1 Feb 90 22:28:08 GMT
Distribution: na
Organization: Advanced Decision Systems, Mountain View, CA  94043, (415) 960-7300
Lines: 86

In response to an update event I'm trying to use CopyBits to blast an 
offscreen pixmap into a window that has an associated palette.  I'm going 
to first describe what I've got, what works, what doesn't work, and then 
ask a few questions.  

Disclaimer:  I try to describe what I think is going on at various points; 
these are my *guesses*!  Do not assume that they are correct!

Ingredients-

First, I have an offscreen pixmap.  This pixmap has an associated color table
that indicates what RGB color I want for each of the possible index values in
the data area.  In this particular case, the index values run from 0 to 255
because the pixel size is 8.  From now on, I'll refer to this as the pixmap.

Second, I have a small window using a palette and pmExplicit so I can monitor
the device color table.  This appears to always work.  I'll refer to this 
window as the color-watcher.

Third, I have my real, display CWindow.  I have a palette associated with this
window that is constructed from the same color table that is used by the
offscreen pixmap above.  The result of this is that I should be able to
say PmForeColor(x) and draw something in my window and the resulting color 
should match (within the palette tolerance) the color in the offscreen
pixmap for index value x.  I'll refer to this window as the window.

What Works -

If I set the pmUsage on the window palette to pmTolerant, and set the 
tolerance to 0 (i.e. exact match only) the following events appear
to happen:
1)  The window gets an activate event.  This causes ActivatePalette() to be
    called which causes the palette manager to do its magic to make sure the 
    colors in the palette are available.  An interesting result of this is
    that the color_watcher, just for a moment, shows that the device color
    table indices contain the same colors as the palette, and in the same
    order.  (I can tell because the colors in the palette are in a different
    order than the standard Apple color map).  They colors in the device then
    seem to scramble around; see question #1.
2)  I do the CopyBits from my offscreen pixmap to the window.  This appears
    to work, but see question #2.

What Doesn't Work (and possibly why) -

If I set the pmUsage to pmAnimated, (so I can animate the colormap), the
following events occur:
1)  The window gets an activate event.  ActivatePalette() grabs the colors
    that I need.  Part of grabbing animated colors involves making those 
    indices unavailable to other windows.  By the time ActivatePalette() has 
    grabbed all the indices that I want, the only ones that are left available 
    to other windows are the ones for black and white.  The color_watcher
    shows the same colors as in the pmTolerant case above, then goes to
    black and white.
2)  CopyBits from pixmap to window.  Result?  A black and white image and a
    black and white color_watcher.  Possible culprit?  CopyBits uses the
    color table of the gDevice of the destination, since it (appears) to need
    an inverse map.  Unfortunately, after the PaletteManager has grabbed
    its colors, the gDevice colormap is only left with black and white, so
    everything ends up black and white.  See (obvious) question #3.

Questions -

1)  After the Palette Manager has given me the colors I want, does it then try
    to re-order the device table so that the difference between my colors and 
    the colors that other windows are expecting is minimized?  After the 
    shuffle described above, the resulting device color table looks similiar
    to the original Apple color map, with my closest color going where the
    equivalent Apple color used to be.
1a) Is the PaletteManager smart enough to know what colors my palette is
    currently not actually using and re-use those indices for other windows?
2)  Does CopyBits try to preserve absolute colors across pixmaps?  Stated
    another way, does CopyBits do something like the following: Grab a pixel 
    from the source, find the absolute color for that pixel by looking at the 
    source color table, look at the (inverse) destination color table to find 
    the index of the color that most closely matches the absolute color, stuff 
    the index that was found into the destination pixmap.
3)  How do you use CopyBits with an animated palette?

Thanks for any clarification and answers!!

Jeff

-- 
Jeff De Vries                        Advanced Decision Systems
Sr. Software Engineer                1500 Plymouth Street
jdevries@ads.com;  (415) 960-7300    Mountain View, CA   94043-1230