🚀 go-pugleaf

In article <1992Nov9.121317.3914@syacus.acus.oz.au> ian@syacus.acus.oz.au (Ian Joyner) writes:
>hrubin@pop.stat.purdue.edu (Herman Rubin) writes:

>>Computer hardware quite properly uses pointers, and any software that
>>does not is just making life difficult for the users.

>Firstly, computer hardware does not necessarily use pointers.

>Secondly, where different architectures do use pointers, the definition
>of them can be quite different. As I have already said, Unisys A Series
>have quite a different kind of pointer to that which C likes. In fact
>the hardware defines a higher level more problem oriented kind of pointer
>than C.

>Thirdly, if you want direct access to hardware capabilities, then you may
>as well use assembler. The disadvantages should be well known. Such languages
>are solutions or 'how' oriented, not problem and 'what' oriented. This means
>that you are not implementation independent, which makes portability difficult.
>This also makes maintainability difficult. This style of programming locks
>you into certain ways of doing things, instead of giving you the freedom
>that higher level languages give you.

It is very difficult to be completely implementation independent, but I
agree it is a good idea to be partially portable and partially implementation
independent.  The current HLLs clearly do not have enough in them to do this;
a very simple example is the process of packing or unpacking floats, which
is mainly portable in use, but not in design.  The infamous 4.2 BSD frexp(),
which used portable code to do something weaker, and often took a long time
to do something simple (it could have even gone on forever), is an example.

In addition, assemblers have atrocious syntax.  At least provide the 
programmer with a good macro processor (a totally non-optimizing mini
compiler, which can use types, storage locations, etc.) to get around 
this.  Even in assemblers which appear to have the same syntax, they
do not; the order of the arguments is not constant for a given instruction.

Portability of certain ends does not mean portability of the means.

>Customers are becoming more sophisticated, and do not like to be locked
>into closed systems (proprietary systems are not necessarily closed), or
>into Unix for that matter. 

But is not locking into a HLL being locked into a closed system?  

Which languages allow the introduction of new operators?  Syntax is
important for both producing code and understanding it.
-- 
Herman Rubin, Dept. of Statistics, Purdue Univ., West Lafayette IN47907-1399
Phone: (317)494-6054
hrubin@snap.stat.purdue.edu (Internet, bitnet)  
{purdue,pur-ee}!snap.stat!hrubin(UUCP)