[Date Prev][Date Next][Thread Prev][Thread Next][Author Index][Date Index][Thread Index]

Re: The Triad's Wild Card

I had three points I wanted to make, but I see Markm made two of them, to wit:

 - It ain't the hyper-magic data structure yet.  It's something else built
   with the same tools (to form the foundation) .  So don't expect speed, and
   reload it after each demo.

 - Because it's slow, and because we want it to look good, it should run
   on the fastest platform we can manage.

Third point:

 - As of a couple nights ago (see my previous letter "Hackin' away..."),
   I've done the necessary minor tweaks to get natCC2 running on the sun3s.
    - These are installed on fester.

    - They differ from what we put on the first tape for Autodesk in two ways:

       - .../nat2.0/sun3/CC2 has been edited.  (The "-B" flag was added
	 to each invocation of cpp.)

       - The native sun3 file /usr/include/sun3/setjmp.h has been edited.
	 (Two "#pragma" lines were commented out, because the compiler
	  didn't understand them AND reported the resulting errors as
	  coming from a DIFFERENT FILE.)

   With these changes in place, proper C++ and X++ code written for sun3
   should compile unmodified on sun4 and vice-versa.  (Regression tests
   may fail wherever they report on pointer values, because of a still
   undiagnosed machine dependency in the memory allocation routines.
   So far, everything else has worked identically.)

   This will allow Greg to work on a sun3 while waiting for delivery of a