Ron Johnson wrote:
Well yeah... but we tried very hard to only add new instructions, not change ones once created. As I recall, there was a family of extended op codes where all the new instructions went.On 06/19/2010 12:00 AM, Miles Fidelman wrote:Ron Johnson wrote:On 06/18/2010 11:05 PM, Miles Fidelman wrote:Right after that, started hacking on MIT's PDP-1 (of Tech Model RailroadClub and Spacewar fame, but at that point free-standing). The really neat thing about the machine was that hackers were allowed, even encouraged, to make HARDWARE changes (e.g., wire-wrapping new instruction codes into the thing). Lots of fun.That sure plays hell on binary compatibility...Hmmm... isn't compatability why we have compilers? :-)But if the opcodes constantly change, you need to modify your compiler every time someone makes a h/w change. And you're screwed if someone replaces the opcode your spiffy program replies on.
Call me old fashioned, but I prefer true high level languages. If you want serious compatability, use Common Lisp :-)Nah, COBOL is where it's at for compatibility.
You're probably right. -- In theory, there is no difference between theory and practice. In<fnord> practice, there is. .... Yogi Berra