xerox Status Report on Alto Lisp May75 free download

File name Status_Report_on_Alto_Lisp_May75.pdf

In tor-Office l\Iemoranclum To CSL, SSL Date May ILl, 1975 From P. Deutsch Location Palo Alto Subject Status report on Alto Lisp Organization PAnC/CSL )(EROX File: (LPD)alispreport.momo (Bravo format) Keywords: Alto, Lisp There has been some recent confusion about the status of the Alto Lisp project. This memo attempts to dispel that confusion by presenting a brief history of the project, a summary of its present state, and our current plans "for further work. The project began in the Fall of 1872 as an outgrowth of some preliminary work on the design of an instruction set which would be tailored to compact representation and rapid execution of Lisp programs: this work resulted from, and the results encouraged, a long-standing interest of mine in Lisp systems for small machines. We decided to proceed by first constructing a system with only an S-expression interpreter (i.e. not using a new instruction set) by simply making a copy of the PDP-I0 Interlisp implementation, converting it from tlfACRO-lO to BCPL and running it on a Nova until Altos became avai lable; then adding the ability to run code which had been compiled into the new instruction set, but emulating the instruction set with a BCPL program; then writing microcode for the Alto to perform the emulation at high speed. The first instruction set design was completed in the Spring of 1973 and presented at the 3IJCAI conference at Stanford in August 1973. In early 1975 we successfully completed the last stage of microcode implementation. It had become clear, meanwhile, that core space would be at a great premium, and that the best way ~o gain space for paging buffers would be to reduce the amount of BCPL code. Especially in view of the fact that Lisp code using the specialized instruction set and a microcoded emulator executes at speeds comparable to BCPL code, we decided to embark on a major program of simplifying the BCPL implementation code and replacing it by Lisp code. In this process we W 0 u I d a p pro a chill U c h m0 r e c los ely the ide a 0 f a Lis p mac h i ne" \'t" h i c h II was entirely self-$ufficient (i.e. did not depend on a Nova/BCPL emulator), gain flexibility which might allow us to incorporate some of the interesting ideas from Smalltalk and Mesa, and make the system more easily maintainable. We are currently pursuing this effort. Our first major step, to replace the original S-expression interpreter (EVAL and APPLY) by compiled Lisp code, is nearing completion. Throughout the course of the Alto Lisp project there has been