After much bad coffee, beating head against keyboard/steering wheel/pillow,
& other general annoyance, Booter 1.9.3 BETA 3 is now available at
<URL:ftp://puma.macbsd.com/pub/NetBSD/booter/Booter1.9.3b3.sit.bin>

with sources at
<URL:ftp://puma.macbsd.com/pub/NetBSD/booter/Booter1.9.3b3-src.sit.bin>

The pertinent portion of the ChangeLog is attached.

The video parameters code is completely new this release, so widespread
testing is encouraged. Hopefully it should be saner than the old Classic
Quickdraw-based routines.

Sorry, but I've not had time to try compiling this with THINK C...
if someone wants to try and drop me a line, that'd be cool...

Enjoy,
-Brian

* Version 1.9.3 BETA 3 -- Mostly cleanup this time around, with 1 important
  bug fix in dialog code and a whole new way to get the parameters of the main
  screen using the Graphics Device Manager and Color QuickDraw routines.

  Dialogs.c: Fixed bug in GroupSet() which was not taking into account argument
  promotion when retrieving item numbers off the vararg stack.

  main.c: Now we check for availability of 68020+68851 or 68030 or better at
  booter startup time. getRAMSize() and MacOS32BitMapping() moved to machdep.c.

  machdep.c: (added.) Most of the boolean functions that result from
  interpreting Gestalt results have been moved here. getRAMSize is now
  GetRAMSize, for consistency.

  ufs_test.c: Changes to include files. load, lowram and mach are history, as
  well as many other superfluous variables of which I've lost track. BootError
  now works like printf(). New max_avail_mem() which should probably migrate to
  machdep.c. Caches are now flushed and disabled before the call to copycode(),
  only. kern_geteof() is history. In copyunix(): Now, load address is LOAD_ADDR
  (a #define.) All the video info comes out of GetMainDevice() and its
  subsidiary data structures; the old code is still there (but #if'd out.) Lots
  more comments, including a try at describing the format of the kernel buffer,
  and documentation of the new algorithms for finding screen parameters.
  Figured out exactly what /* LAK says I need this. */ meant: copycode() must
  be copied to the end of the buffer, which must end on a page boundary. The
  first two lines round down len to the next lowest page boundary, and then
  harry is assigned to be 1KB below that.

--
Brian R. Gaeke <brg@dgate.org> - Ask me for my PGP public key!
"If I were a blue spider, I would certainly ride on a train all the way from
Avallon to Paris, and I would set up my house on the nose of a chocolate
penguin. It's just a matter of common sense." James Wright, "Against Surrealism"