[SCC_Active_Members] APL

[Dennis Allison]

John Walters was at IBM where he did the 360 Assemblers.  Last heard of he 
was a Professor in Sweden (Linköpings universitet) about half the year.  
He described the HP APL compiler in an EE380 talk -- we have the tape 
somewhere but I don't remember the actual date.

Jim Ryan did a similar system for the Burrooughs 6500, actually a layered
interpreter which used several different represenations and profiling to 
decide which one to use.  He described this in detail at a SIG meeting in 
Santa Cruz.



On Wed, 28 Sep 2005, Bob Fraley wrote:

> Hi,
> 
> I heard that there is some investigation into APL.  HP's APL was mentioned 
> in the discussion.  I've just made contact with Ken Van Bree, one of the 
> programmers, and Alan Marcum, who was an intern from MIT who did some work 
> on the project.  I don't know how to contact most of the people who were on 
> the project, but here are some of the names:
> 
> Paul Stoft:  Headed HP Labs, and sponsored this project.
> 
> John Walters:  Project manager (formerly of IBM?)
> 
> Developers:
> Ron Johnson
> Grant Munsey
> Eric Van Dyke
> Ken Van Bree
> 
> Rob Kelly:  Stanford student who was on the project
> Alan Marcum:  MIT student intern on the project
> 
> All of the above people were in the research labs.
> 
> Jean Danver and Bob Crum:  helped take the project into production.
> 
> For those who have not heard much about this project, this implementation 
> of the APL system was essentially a JIT system.  It compiled the APL code 
> to speed up execution.  Normally, APL is interpreted because there is no 
> data typing.  Each entry to a procedure may have parameters having 
> different types, and these types include multidimensional arrays.  Each 
> time the procedure is called, one of the existing compiled versions  will 
> be used if the types match.  Otherwise, a new version of the procedure will 
> be compiled with the new parameter signature.  (Smaller code elements could 
> also be compiled.)
> 
> The notable piece of hardware was a version of the HP terminal that had an 
> APL keyboard and display.  There might have been some changes to the 
> processor to speed up certain operations; I'd have to inquire further to 
> find that out for sure.
> 
> This is all explained in an HP Journal article.  I don't know if the HP 
> Library has archived these journals.
> 
> If anyone would like contact information, I believe that we can get in 
> touch with 3 or 4 of these people.
> 
> Bob
> 
> 
> _______________________________________________
> SCC_active mailing list
> SCC_active at computerhistory.org
> http://mail.computerhistory.org/mailman/listinfo/scc_active
> 

--