mirror of
https://git.sr.ht/~crc_/retroforth
synced 2024-11-16 19:48:56 +01:00
4c53181624
FossilOrigin-Name: a7e5fe4c71047a01a833ce31583ef5597f3f0235c5a2fab90e55f07510bc0bf5
51 lines
1.3 KiB
Forth
51 lines
1.3 KiB
Forth
# Accumulator
|
|
|
|
## Description
|
|
|
|
This implements a function that takes an initial value and constructs a
|
|
new function that returns the value before incrementing the stored value
|
|
by 1.
|
|
|
|
So, given an initial value of 1, the first time the function is called,
|
|
1 is returned. The second, 2, and so on.
|
|
|
|
In traditional Forth, this would be done using a CREATE/DOES> construct.
|
|
RETRO allows for something similar using the `does` combinator.
|
|
|
|
An example in a traditional Forth:
|
|
|
|
: acc ( n "name" -- )
|
|
create , does> dup >r @ dup 1+ r> ! ;
|
|
|
|
In RETRO, we could begin by rewriting this using the RETRO words:
|
|
|
|
:acc (ns-)
|
|
d:create , [ dup push fetch n:inc pop store ] does ;
|
|
|
|
The `dup push ... pop` pattern is the `sip` combinator, so we can
|
|
simplify it:
|
|
|
|
:acc (ns-)
|
|
d:create , [ [ fetch n:inc ] sip store ] does ;
|
|
|
|
This is better, but not quite done. RETRO has a `v:inc` for incrementing
|
|
variables, which would eliminate the n:inc and store. And a `bi`
|
|
combinator to run two quotes against a value. So we could simplify yet
|
|
again, resulting in:
|
|
|
|
~~~
|
|
:acc (ns-)
|
|
d:create , [ [ fetch ] [ v:inc ] bi ] does ;
|
|
~~~
|
|
|
|
This removes the primitive stack shuffling, and leaves something that
|
|
expresses the intent more clearly.
|
|
|
|
Finally, here's a little test case:
|
|
|
|
```
|
|
#10 'foo acc
|
|
foo
|
|
foo
|
|
foo
|
|
```
|