[Baypiggies] M.I.T. to drop Scheme in favor of Python

Roderick Llewellyn roderick at sanfransystems.com
Sat May 30 18:09:06 CEST 2009


When I was at MIT, the most popular language was probably PL/1. This was a typically "ibmic" language, it did it all. Was it a scientific language? Was it a business language? Was it a string-processing language? Yes.
Assembly languages were also popular. There were three main computing facilities: Multics, PDP/10 used in AI type research, and IBM mainframes. I wrote a lot of PL/1 on Multics, and co-authored the Multics APL system sold by Honeywell. APL was also popular. I wrote many thousands of lines of IBM assembly language. On the PDP-10s of course Lisp, as well as its assembler, was widely used. I don't think course 6.001 was around when I was there; MIT didn't really have an introductory programming course. They assumed you already knew how to program. A major early CS course was Donovan's 6.251 on operating systems which made use of IBM360 assembler and PL/1.
My first job after graduation was with Digital Equipment Corp (ironic because I hadn't used the DEC equipment while at MIT). Everything there, from OSes to text editors to newspaper typesetting systems, was written in assembly language. I was in a group that tried to bring "structured programming" concepts to DEC. There was a lot of pushback from engineers who liked the efficiency of assembler. A typical programming situation was that there was a routine that people used, call it X. Now to save some memory, code Y would not call X, it would directly jump to X (i.e., w/o saving a return address on the stack), knowing that routine X would jump back to Y when done. Or sometimes the return address was put in a register. (This was also common on IBM mainframes, in which there was an instruction BALR R1,R2 which would jump to the address held in register R2 while storing the address of the next instruction following the BALR in register R1). Later on Z would be written and would want to call X, so now the code would need modifying to use the stack for saving return addresses. But nobody could easily find all the places to fix up, and bugs accumulated.... I saw one piece of code that pushed 10 or so extra copies of the return address just in case!

By the way it was fun meeting everybody Thursday nite, jj, Aaron, Glen, Daryl, Wesley, Vicky, etc. I hope to see you all again soon.
- Roderick Llewellyn
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mail.python.org/pipermail/baypiggies/attachments/20090530/b87ad6cf/attachment.htm>


More information about the Baypiggies mailing list