Another Concatenative Language -- dc
Jack Waugh — 2001-08-06 00:32:40
On any Unix-like system, do a "man dc". 'dc' (desk calculator) has a
stack and a namespace (with only single-character names). A single
character denotes each function (for example, + adds the top two
stack elements and replaces them with the result; d duplicates the
top element; p prints the top element without consuming it; x
executes the top element as a program). Quoting character strings
suffices to quote code. There are combinators. Arithmetic runs to
arbitrary precision. dc is not as purely functional as Joy (ignoring
I/O) is, because of the namespace operators. In fact the only way to
pop the top element and discard it is to store it under a name (I
use "g" for garbage, so the function is "sg", store in garbage) (I
lied about functions being always denoted by just one character;
the "store" and "load" functions take two characters: "s" or "l"
followed by the name of the cell in the namespace).
Manfred von Thun — 2001-08-06 02:06:35
On Mon, 6 Aug 2001, Jack Waugh wrote:
> On any Unix-like system, do a "man dc". 'dc' (desk calculator) has a
> stack and a namespace (with only single-character names). A single
> character denotes each function (for example, + adds the top two
> stack elements and replaces them with the result; d duplicates the
> top element;
You are right, I had forgotten about dc's dup operator d.
So what I wrote on the Joy FAQ just three days ago is wrong:
According to the definition I use there,
dc is not postfix but concatenative. I'll amend the FAQ.
> p prints the top element without consuming it; x
> executes the top element as a program). Quoting character strings
> suffices to quote code. There are combinators. Arithmetic runs to
> arbitrary precision. dc is not as purely functional as Joy (ignoring
> I/O) is, because of the namespace operators. In fact the only way to
> pop the top element and discard it is to store it under a name (I
> use "g" for garbage, so the function is "sg", store in garbage)
Indeed, so we have dup and pop
> (I
> lied about functions being always denoted by just one character;
> the "store" and "load" functions take two characters: "s" or "l"
> followed by the name of the cell in the namespace).
And swap can be done by sa sb la lb
So the three principal stack operators can be effected in dc
Thanks for pointing it out.
Being envious of the infinite precision in dc (and bc),
some time ago I even considered having a Joy package which
uses dc to do the hard work, hidden away from the user by an
invisible "system" call. But I rejected the idea for the time being.
Maybe somebody should look at it again.
- Manfred
John Cowan — 2001-08-06 03:31:48
Manfred von Thun scripsit:
> Being envious of the infinite precision in dc (and bc),
> some time ago I even considered having a Joy package which
> uses dc to do the hard work, hidden away from the user by an
> invisible "system" call. But I rejected the idea for the time being.
> Maybe somebody should look at it again.
There are arbitrary-precision packages written in C. I'll look into
this at some point, though bignums aren't something which really
fascinate me.
--
John Cowan
cowan@...
One art/there is/no less/no more/All things/to do/with sparks/galore
--Douglas Hofstadter
John Carter — 2001-08-06 03:53:25
On Mon, 6 Aug 2001, Jack Waugh wrote:
> On any Unix-like system, do a "man dc".
In that realm is the Hp48 calculators RPL (Reverse Polish Lisp).
In the opposite realm there is a converter that translates a lispish
version of postscript into ordinary postscript!
Hp48 RPL was what got me interested in postfix based languages in the
first place.
I suddenly realised, while programming my calculator, that the huge fuss
made about syntax is totally irrelevant compared to the semantics. ie. A
postfix language is almost syntaxless, all you need is a lexical scanner,
no parser needed.
Unfortunately Computer Science departments have given a generation of
students an obsessive fixation of syntax and parsers and parser
generators. All that effort and thought has been misdirected.
What would have happened if all that creativity had been expended on the
semantics instead of increasingly complex syntaxes that render reflection
impossible....
John Carter Phone : (64)(3) 358 6639
Tait Electronics Fax : (64)(3) 359 4632
PO Box 1645 Christchurch Email :
john.carter@...
New Zealand
I'm becoming less and less convinced of humans as rational beings.
I suspect we are merely meme collectors, and the reason meme is only
kept on to help count our change.
Jack Waugh — 2001-08-06 16:24:04
Oops; dc doesn't have a string concatenation operator. It provides
no programmatic way to build programs.
In fact, there seem to be no ways to build any kind of aggregate,
unless the : and ; operators do something like that (the doc says dc
provides them for bc's array operations but doesn't say what they do).
dc's program quoting has the same syntax as that of Joy: square
brackets (Smalltalk and Self also quote code with square brackets).
Too bad there's no cons.
Manfred von Thun — 2001-08-07 00:15:10
On Sun, 5 Aug 2001, John Cowan wrote:
> There are arbitrary-precision packages written in C. I'll look into
> this at some point, though bignums aren't something which really
> fascinate me.
Thanks.
No urgency at all, I think. There is also the possibility that
those packages are not widely portable, but I do not know
anything about them.
- Manfred
Manfred von Thun — 2001-08-07 00:39:37
On Mon, 6 Aug 2001, John Carter wrote:
[deletions]
> Hp48 RPL was what got me interested in postfix based languages in the
> first place.
>
> I suddenly realised, while programming my calculator, that the huge fuss
> made about syntax is totally irrelevant compared to the semantics. ie. A
> postfix language is almost syntaxless, all you need is a lexical scanner,
> no parser needed.
>
> Unfortunately Computer Science departments have given a generation of
> students an obsessive fixation of syntax and parsers and parser
> generators. All that effort and thought has been misdirected.
Agree, at least in part. I consider myself fortunate in having only
very few books (on Pascal) in my early days. One was the Jensen-Wirth
Pascal manual, the other Wirth's Algorithms + Data Structures = Programs,
the first, 1976 edition which had PLZERO. I learnt all about recursive
descent parsing and would always design a language so that this is
possible, especially the ease of error reporting and recovery.
But I do think that the idea of syntax directed parsing and translating
is something everyone should learn, and looking inside the various
Pascal compilers (PL0, PascalS, P4) is useful.
> What would have happened if all that creativity had been expended on the
> semantics instead of increasingly complex syntaxes that render reflection
> impossible....
After some experience with Pascal I did study Lisp a bit, I think it
was Anatomy of Lisp or something like that. Yes, Program = Data,
wonderful stuff. Joy eventuated essentially out of that..
- Manfred
Manfred von Thun — 2001-08-07 01:35:19
On Mon, 6 Aug 2001, Jack Waugh wrote:
> Oops; dc doesn't have a string concatenation operator. It provides
> no programmatic way to build programs.
The "Oops" leads me to think that you have this argument in mind:
dc does not have a concatenation operator
---
therefore, dc is not a concatenative language
If this is what you meant, then read on, else ignore:
The way I define things is this:
A language is concatenative iff
for all programs P and Q
if P denotes a function F and Q denotes a function G,
then their concatenation P Q denotes the composition of F and G
This is quite independent of what sort of data structures
the language provides. It may have just integers and nothing else.
If it also has quoted programs, used by combinators, then
a concatenation operation would be useful because it makes
the language reflective. But the concatenation operation
_in_the_language_ is not a defining characteristic of concatenative
languages.
- Manfred
Manfred von Thun — 2001-08-08 05:08:27
Unfortunately I lost the original mail with the question from John Carter:
Why is == an infix operator?
The question has been raised before, including at least once on this
mailing group. Since it recurs, I should really start another
"Further Frequently Asked Questions", and I'll make that question
a priority. But it will take at least a week, I'm a little snowed
under. Sorry.
I was able to read the details of your question by looking back
at yahoo's archive. The way postfix does it sounds interesting.
But I would like to see some examples of where those extra
facilities are really useful.
- Manfred
waugh@acm.org — 2001-08-08 13:17:08
> Unfortunately I lost the original mail with the question from John
Carter:
. . .
Loosing "mail" that came through this forum need never be a problem.
Just go to the message archive. In fact, I recommend not receiving
mail. The message archive resides at
http://groups.yahoo.com/group/concatenative/messages