🚀 go-pugleaf

In article <INFO-M2%90070316593556@UCF1VM.BITNET> Modula2 List <INFO-M2%UCF1VM.BITNET@PSUVM.PSU.EDU> writes:
>Anyone else have any answers as to why Wirth designed such a restriction?

Perhaps Randy Bush, or some ETH folk would care to comment on the following:


Although one is perhaps overextended to speak for Wirth, I have at least
read several of his papers, and --whether he intends or not-- I consistently
come up with a few design philosophies in his work.

1)  His papers have the 'flavor' or a mathemeticians writing.  My BA degree
is in math.  So...
   a)  A project should be complete for the original job/specification in
       mind.
   b)  A project should be minimal.
       'complete' and 'minimal' attain special meaning to a mathemetician.
       in lay terms, they keep their usual meaning, but it's pretty strong.

Everything else seems subordinate to this.

2)  The program is a document (formalized to a 'Text' now, in Oberon).
    a)  Excessive obscureness, for performance, typically can be avoided
        through better algorithm abstraction, and failing that, better
        hardware.  Failing that, N.W. seems able to design a new language.
        
        Before I carelessly start a flame war, let me point this out about
        my background.  Once, in 1974, a Ms Comp. Sci major spent 4 months
        working in a 4Kb RAM environment trying to squeese 13 bytes out of
        a piece of code.  Memory was about $1,000 / Kb, and a month's house
        payment was $85.  I still tend to 'Hack,' but after studying N.W.'s
        papers, I leave it to the week before delivery, and my bug rate is
        fairly low.

        Anyway, I think this point focuses the preferences between the
        Pascal vs. C communities.

    b)  Textual locality of reference actually improves program comprehension.

3)  It is permissable to design the language, in part, to make the software
    development environment 'good.'

    a)  Fast Compilation --> Simple language.
    b)  Flexible or Custom Environment --> Libraries external to language.
    c)  Correct products --> Safe language constructs, simple compilers.

4)  Compiler optimization doesn't strongly interest him.  At a talk, I
    heard him say that his one pass Modula-2 compiler used 20 year old
    compiler technology, he knew that, and one can only do so much with
    a handful of graduate students.  Also, he believes that the language
    design itself influences the quality of code, and that optimization of
    a cumbersom language raises the risk of an incorrect compiler.


---------------------

If these abstracts really represent(ed) Wirth's thinking, then I think the
business of the FOR loop is kept simple, the references local, the optimization
is built in (no assignments are legal to the loop variable within the
loop:  It is safe to put the value into any special place, such as registers,
without execution time safeguards.


-- 
Aubrey McIntosh  	"Find hungry samurai." -- The Old Man        
1502 Devon Circle       comp.os.minix, comp.lang.modula2         
Austin, TX 78723 
1-(512)-452-1540  (v)