rgb888 => full-24-bit-color colorindex (like rgb565)
vector picking
vecalpha command
hot-key help ("?")
snapshot [-q jpeg_quality]
snapshot ... @  e.g. snapshot fred.@.%04d.png  => stereo-pair snapshot
pixelaspect command
don't crash in snapshot() if window has evaporated
insist on using GLEW library

If a color-index variable is named rgb888, it's interpreted as a full 24-bit color value:
  R*65536 + G*256 + B
with 0 <= R,G,B <= 255.

parti_snapshot(): if the pattern includes '@', emit a stereo pair of images, replacing @ with L and R.

   ?      print help message for hot keys
   ctrl-S toggle stereo left/right eyes
Safer sprintf for subcam report.

Use static GLEW library under MinGW (libglew32.a rather than libglew32.dll.a).

Use static GLEW library under mingw (libglew32.a rather than libglew32.dll.a).

Revamp FLTK configuration -- use fltk-config.  It tells us which libraries we need.  It also can tell us where to find libpng and libjpeg - either the versions built-in to fltk since 1.1.x, or else the system version if available.

Regenerate from (unchanged) partipanel.fl with newer version of fluid.   Switches from FL_Input to FL_Float_Input.

Make use of vecalpha value.

Add some redraw()s when we detach opengl window from GUI, to avoid MacOS Lion bug which left them un-refreshed.

Add some redraw() calls to avoid MacOS Lion bug that left cmd box, cmdhist scrollback, and opengl view to go un-refreshed.  Especially when we switch on the timestep row.

Remember to schedule redraw().   Probably not necessary, but trying to squash bug that's appeared in MacOS Lion.

Check the right thing -- ac_cv_header_jpeglib_h -- to see whether we've found the JPEG header and need to find corresponding library.

For MacOS (Lion), use -framework Cocoa rather than -framework Carbon, following fltk-1.3's lead.  No, I don't understand why.

so it doesn't collide with VERSION file.

Try to make jpeglib.h work right under Windows/MinGW.  Need to declare our own "boolean" type for some stupid reason.

Default jpeg quality = 97.

Flip Y when encoding PNGs.
Always free the snapshot buffer.

Use hack-y MINGW32 test to see which version of GLEW library to use --
"-lGLEW" on most systems, "-lglew32.dll" (libglew32.dll.a) on MinGW.

   - four digits by default
   - use .png suffix if libpng available, or if unix (assumes "convert" available)
   - use .ppm suffix if WIN32 and no libpng

for comparison with -c 'early commands', before window setup.