Hello!
Why is it that I can't just store any with-... combinator in a SYMBOL, then
use `SYMBOL get call`?
I get it that we want to check the stack effects, but doesn't it break the
idea of concatenativity?
The fact that we inline all the with-... combinators -- is there a way around
it?
You'll want a use case, so I'll make one up.
Let's say we have a hypothetical word, call it parse-lines-interactive:
: parse-lines-interactive ( lines -- quot/f )
[ [ parse-lines ] with-ctrl-break
] with-compilation-unit ;
And let's say we have another completely made up piece of code that does this:
read-quot [
'[ [ datastack _ with-datastack ] with-ctrl-break ]
[ call-error-hook datastack ]
recover
] [ return ] if*
In both cases we use the `with-ctrl-break` combinator, which is `inline`.
Let's say we have two implementations of it, which we want to dynamically
switch. We begin by creating the two implementations and a SYMBOL like so:
: with-ctrl-break ( quot -- )
enable-ctrl-break
[ disable-ctrl-break ] [ ] cleanup ; inline
: with-no-ctrl-break ( quot -- )
call( -- ) ; inline
SYMBOL: break-handler
[ with-no-ctrl-break ] break-handler set-global
You can see that both `with-ctrl-break` and `with-no-ctrl-break` have the
same stack effect: they consume one quotation.
Now we update the code to dispatch on the SYMBOL:
: parse-lines-interactive ( lines -- quot/f )
[ [ parse-lines ] break-handler get call( quot -- )
] with-compilation-unit ;
read-quot [
'[ [ datastack _ with-datastack ] break-handler get call( quot -- )
]
[ call-error-hook datastack ]
recover
] [ return ] if*
Amazingly, this no longer works! Now, I'm asking you: where's the
concatenativity in that?
---=====---
Александр
------------------------------------------------------------------------------
_______________________________________________
Factor-talk mailing list
Factor-talk@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/factor-talk
