various questions

peter_easthope@gulfnet.pinc.com — 2000-05-11 15:55:49

Sorren, why are you creating a new language and 

writing the compiler?  If you have explained already 

perhaps you can just point me back to the message.



Have any of you folks used J?  The concatination here

is reminiscent of J.



wtanksley@... said,

wt> the syntax I used depended on strings, while 

yours depends on "symbols". 



I know what a string is.  What is a "symbol"?



Brent Kirby mentioned,

bk> Count upward, beginning at 0, skipping multiples of five



Sorren replied,

sr> 0 [dup 5 % 0 = [dup puti] [] if 1 +] spin



In case anyone is interested, this is it in J without division.

J evaluates each line from right to left. (OK, it wasn't my 

choice.)



screen=. (1!:2)&2  NB. Display the argument as a side effect;

NB. equivalent to puti.

si=. screen@>:     NB. screen after increment 

count=. (>:@si@si@si@si)^:_

count 0



^: means "apply the function on the left the number of times

given on the right".  _ is "infinity".  Hence ^:_ is equivalent to 

your spin.  The four lines above can be condensed to these two.



si=. (1!:2)&2@>: 

((>:@si@si@si@si)^: _) 0





wt> ... more thinking about continuations. 



What is a continuation?



Thanks,       Peter E.

srenner@mail.ru — 2000-05-11 17:39:05

--- In concatenative@egroups.com, peter_easthope@g... wrote:

 > Sorren, why are you creating a new language and 

> writing the compiler?  If you have explained already 

> perhaps you can just point me back to the message.

> 

> Have any of you folks used J?  The concatination here

> is reminiscent of J.

> 

> wtanksley@b... said,

> wt> the syntax I used depended on strings, while 

> yours depends on "symbols". 

> 

> I know what a string is.  What is a "symbol"?

> 

> Brent Kirby mentioned,

> bk> Count upward, beginning at 0, skipping multiples of five

> 

> Sorren replied,

> sr> 0 [dup 5 % 0 = [dup puti] [] if 1 +] spin





 
Well, I DIDN'T write that last line. It was either iepos or tanksley.

Probably the bit about writing a new language and compiler isn't

correctly aimed at me either. However, it might be, because I am

implementing Joy in Oberon. So I will answer: "because I am not

intelligent enough to understand how a virtual machine works without

writing one." 



I think that what I wrote yesterday is complete. That is, it will need

some operations and combinators besides dup and i, and a scanner so it

can parse input into programs, and more literal types. Characters and

strings should be easy. Joy also has a set type. Oberon has a set type

too, though I have never used it. Probably one will map to the other.

There will need to be a dictionary mapping names of ops to ops, for

the scanner at least. 



Here is an idea: multiple Joy threads that communicate through a

global store of variables that can be bound only once, like Mozart Oz.

It would mean adding "variables", but only global ones, so each VM

(thread) would be no heavier than it would be in pure Joy.  



sr



"Hitting today's nail with the hammer of tomorrow. On the head.

egrollard.com"