Lambda the Ultimate

Randy Bryant, dean of the school of computer science at CMU, sent out an email saying that John C. Reynolds passed away yesterday.

Subject: In Memoriam. John Reynolds, June 1, 1935 - April 28, 2013
Date: Sun, 28 Apr 2013 21:45:12 -0400
From: Randy Bryant
To: scs-all@cs.cmu.edu

I'm sorry to announce that John Reynolds, a long-time member of our computer science faculty, passed away early this morning. Many of you know that John had been in declining health recently. We were able to celebrate his retirement him last summer. He had a heart attack last week and went downhill over a period of several days.

John got his PhD in 1961 in theoretical physics, but while working at Argonne National Laboratory came to realize that his passion was for computation. He became a very successful computer scientists, focusing on the logical foundations of programs and programming languages. He was at Syracuse University from 1970 to 1986 and then joined the CSD faculty.

John has made many important contributions over his career. Interestingly, his 2002 work on separation logic, done jointly with Peter O'Hearn and others, has been especially prominent. Separation logic provides a formal way to reason about what we might think of as "normal programs," i.e., ones that operate by changing the values stored in memory, but where memory is partitioned into independent blocks, and so we can reason about different program components independently. I can only hope that the work I do at age 67 would be counted among my best!

We will also remember John for this cheerful spirit, his high ethical standards, and his deep intellect. He will very much be missed.

Randy Bryant

It's probably impossible to overstate the impact that John had on the field of programming languages. But beyond being a great scholar, he was also a generous mentor and a fundamentally decent and kind human being. He will indeed very much be missed.

Intuitively, the semantics of STM is appealingly simple. However, as researchers are coming to discover, this simplicity is illusory and the actual semantics offered by implementations are often counterintuitive—programs that look “obviously correct” may behave in unexpected ways.

to be seen in C# and maybe Python.

I've been busy developing the SPREAD language. After a lot of tinkering, I'm pretty much done with the semantics. I believe the result can be considered 'kind of weird' but very powerful.

I have a hard time classifying the SPREAD language - may be you can help me classify it? The best classification I've come up with is that SPREAD is 'a retro-active programming language'.

Bold claim: retro-active programming subsumes reactive functional programming, and much more.

Comments are very much appreciated!