Clean strictness analyser

[rinus plasmeijer]

Dear David,


>Does Clean still use Eric Knocker's Abstract Reduction to do strictness
>analysis in the compiler?

Yes!

>I ask because, as a theoretician, I am curious as to whether there is
>still any interest in yet an account of how abstract reduction works.

Of course there is. I hope you are aware that Eric is not working in our=
 team anymore (he is working for the big industry). The SA of Eric is a=
 great piece of work. We are currently working on a new Clean compiler=
 (Clean 2.0), completely rewritten in Clean. Well almost completely: to safe=
 work we will reuse part of the old compiler, namely the code generation=
 part (core Clean to ABC code) written by John van Groningen and the SA=
 written by Eric.
Both very stable pieces of code and very good.

>The account in Eric Knocker's thesis actually dealt with infinitary term
>
>rewriting rather than graph rewriting and he said in the thesis that it
>could not be viewed as complete. The earlier account he did with Marko
>van Eekelen, Chris Hankin and Eric Goubault also had some short-comings
>since they voluntarily excluded graphs with cycles via Union nodes.

True.

>I am interested in producing a (possibly long) account of abstract
>reduction that uses material from Eric Knocker's thesis, and any other
>sources, together with material from my own thesis that to give both an
>easy to understand model and a detailed explanation of how the analyser
>works.
>
>Is this still of interest or should I just "let sleeping dogs lie "?

No, we sre still iterested. But since Eric has left, the SA is working fine=
 and we have lots and lots of new ideas to work out, we did not work further=
 on this part.

Do you looking for someone to collaborate with our are you just wondering=
 whether we are interested in your results?
The latter is certainly the case.
Oyhow, we ceratinly want to help you if you have questions and are of course=
 open for any further collaboration...


Greetings,

Rinus