[Caml-list] camlp4: ONCE construct

[Caml-list] camlp4: ONCE construct

[jehenrik]

Hi,

I would like to write a Camlp4 construct which causes its argument to 
only be evaluated once.  Eg:

...
	...
		...
			ONCE Pcre.regexp "([0-9]+)"

would be transformed to:

let gensym000 = Pcre.regexp "([0-9]+)";;
...
	...
		...
			gensym000


but I don't see a way to get a nonlocal transformation like that.  One 
way would be if the system had a way to side-effect out a new toplevel 
let while the expressions were transforming.  I suspect that facility 
does not exist.  Another way would be to have a second keyword 
(ONCEBLOCK or something?) that would go where the let bindings 
eventually would, which could take as an argument the entire expression 
tree containing ONCE declarations.  Then after a complicated recursive 
parse on all the syntactic constructs, all the ONCEs could be located, 
enumerated, and removed.

Note that I don't want to "memoize," or create a thunk which when 
evaluated the first time does something complicated and all subsequent 
times remembers what it did the first time.  That still requires 
manually writing the let statements in a different code location.  My 
goal is to improve readability by putting my regexps where they matter, 
not several screens up just so that they get cached at runtime.

Thanks for any advice,


Jeff Henrikson

-------------------
To unsubscribe, mail caml-list-request@inria.fr Archives: http://caml.inria.fr
Bug reports: http://caml.inria.fr/bin/caml-bugs FAQ: http://caml.inria.fr/FAQ/
Beginner's list: http://groups.yahoo.com/group/ocaml_beginners

Re: [Caml-list] camlp4: ONCE construct

[Yutaka OIWA]

>> On Fri, 10 Jan 2003 14:14:14 -0500, jehenrik <jehenrik@yahoo.com> said:

jehenrik> Hi,
jehenrik> I would like to write a Camlp4 construct which causes its argument to
jehenrik> only be evaluated once.  Eg:

jehenrik> ...
jehenrik> 	...
jehenrik> 		...
jehenrik> 			ONCE Pcre.regexp "([0-9]+)"

jehenrik> would be transformed to:

jehenrik> let gensym000 = Pcre.regexp "([0-9]+)";;
jehenrik> ...
jehenrik> 	...
jehenrik> 		...
jehenrik> 			gensym000

See an article by Francois Pottier at
http://caml.inria.fr/archives/200107/msg00187.html
for (almost direct) answer.

Here is my question: Is there any way to declare a
module-local variable without using .mli signature?
It may not be prefarable for direct use by OCaml programmers
(than signature approach), but is useful for camlp4 programmers
to avoid name crashes caused by "open" or "include" directive.

P.S.
I have implemented a more powerful syntax extension for regular
expression matching, and I'm now preparing a documentation.
I'll release it for caml-list within this month (hopefully).

-- 
Yutaka Oiwa              Yonezawa Lab., Dept. of Computer Science,
      Graduate School of Information Sci. & Tech., Univ. of Tokyo.
      <oiwa@yl.is.s.u-tokyo.ac.jp>, <yutaka@oiwa.shibuya.tokyo.jp>
PGP fingerprint = C9 8D 5C B8 86 ED D8 07  EA 59 34 D8 F4 65 53 61
-------------------
To unsubscribe, mail caml-list-request@inria.fr Archives: http://caml.inria.fr
Bug reports: http://caml.inria.fr/bin/caml-bugs FAQ: http://caml.inria.fr/FAQ/
Beginner's list: http://groups.yahoo.com/group/ocaml_beginners

Re: [Caml-list] camlp4: ONCE construct

[jehenrik]

> See an article by Francois Pottier at
> http://caml.inria.fr/archives/200107/msg00187.html
> for (almost direct) answer.

Well, for my original question, consider the choice of regular 
expression compilation to be an arbitrary task.  The only point is that 
it be computationally expensive.  I have wanted the same thing in 
completely different contexts.


Jeff Henrikson


jehenrik> Hi,
jehenrik> I would like to write a Camlp4 construct which causes its 
argument to
jehenrik> only be evaluated once.  Eg:

jehenrik> ...
jehenrik> 	...
jehenrik> 		...
jehenrik> 			ONCE Pcre.regexp "([0-9]+)"

jehenrik> would be transformed to:

jehenrik> let gensym000 = Pcre.regexp "([0-9]+)";;
jehenrik> ...
jehenrik> 	...
jehenrik> 		...
jehenrik> 			gensym000

-------------------
To unsubscribe, mail caml-list-request@inria.fr Archives: http://caml.inria.fr
Bug reports: http://caml.inria.fr/bin/caml-bugs FAQ: http://caml.inria.fr/FAQ/
Beginner's list: http://groups.yahoo.com/group/ocaml_beginners