Re: PL/MIX

glen herrmannsfeldt <gah@ugcs.caltech.edu> writes:


> Note that PL/360 really a high level syntax on an assembler, and
> not what most would call a high level language.
> Consider that the PL/360 statement
> R1:= R1+R1+R1;
> will result in R1 being multiplied by four. It compiles to something like:
> LR R1,R1
> AR R1,R1
> AR R1,R1
> requiring the thought process of an assembler programmer.


This is very helpful. For one thing, it helps to define the problem of
what PL/MIX is supposed to be like and that, for me, is the primary
problem at this point.


One of the constraints seems to be that one wants to be able to treat
registers and memory locations as variables and to combine them using
a syntax something like Algol. I myself have never used Algol, but I
have used computer algebra packages such as REDUCE and MACSYMA and
they are described as having an Algol like syntax. REDUCE, in
particular, if I remember correctly, uses := for assignment (MACYMA
uses : for ordinary assignment and := for function definitions).


So, the problem is to define something like a computer algebra package
where all the operations represent assembly language operations and
where the commands are directly translated into the corresponding
assembly language commands (with the caveat Glen Herrmannsfeld
mentions regarding associativity).


Am I missing anything else?


Anyway, this provides me with more motivation for reading the PL/360 manual at
http://www.slac.stanford.edu/spires/explain/manuals/PL360TXT.HTML
--
Ignorantly,
Allan Adler <ara@zurich.csail.mit.edu>

Return to the comp.compilers page.
Search the comp.compilers archives again.