[Caml-list] CamlIDL question

[Caml-list] CamlIDL question

[Yurii A. Rashkovskii]

Gentlemen,

Is the below trick is possible in CamlIDL?

I want to wrap C library that uses structs for storing
pointers to functions, like

typedef struct {
        int (*open)(char *filename);
} blablabla;

I want to wrap it automatically to something like

type blablabla = 
{
   open: (string -> int)
}
;;

Is it possible?


-- 
Thanks in advance,
Yurii.
-------------------
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] CamlIDL question

[Xavier Leroy]

> Is the below trick is possible in CamlIDL?
> I want to wrap C library that uses structs for storing
> pointers to functions

No, CamlIDL doesn't support C function pointers.  The reason is that
mapping C functions pointers to Caml functions is easy, but mapping
Caml function closures to C functions is nearly impossible (there is
nowhere to store and pass the environment part of the closure).

However, CamlIDL support COM interfaces, which are basically a subset
of C++ classes.  These support two-way conversions between OCaml
objects (i.e. more or less collections of Caml functions) and COM objects
(i.e. more or less collections of C function pointers + instance
variables).

- Xavier Leroy
-------------------
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] CamlIDL question

[Dmitry Bely]

Xavier Leroy <xavier.leroy@inria.fr> writes:

>> Is the below trick is possible in CamlIDL?
>> I want to wrap C library that uses structs for storing
>> pointers to functions
>
> No, CamlIDL doesn't support C function pointers.  The reason is that
> mapping C functions pointers to Caml functions is easy, but mapping
> Caml function closures to C functions is nearly impossible (there is
> nowhere to store and pass the environment part of the closure).

It's possible using the "trampoline" technique, when the necessary small
stub functions are generated on the fly in the data (or stack)
segment. This is how gcc's nested functions work. There is "ffcall" library
that implements trampolines for (probably) all hardware platforms where
OCaml may run. Am I using it with OcamlIDL to do just what Youry asked
for. Unfortinately ffcall's author website has been disappeared, but it
should not be a problem to find "ffcall-1.8c.tar.gz" somethere in the
Internet.

- Dmitry Bely


-------------------
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] CamlIDL question

[Chris Hecker]

>There is "ffcall" library
>that implements trampolines for (probably) all hardware platforms where
>OCaml may run. Am I using it with OcamlIDL to do just what Youry asked
>for.

I was planning on using the ffcall library to make dynalinking to C 
functions that take function pointers myself (you can see in the list 
archive a year or so ago).  I'm glad somebody's made it work.

However, the new bee in my bonnet is to have ocaml generate (or have it in 
the asm files for each platform) a sort of meta-routine that can be called 
at runtime to dynamically generate a range of different platform-specific 
functions that are necessary for this kind of lowlevel work.  You want some 
kind of routine that can generate a closure/thunk like ffcall, or switch 
the stack and cpu state like in the fiber/thread stuff, etc.  The idea is 
that since the ocaml runtime knows the platform (since it was compiled), it 
makes sense to have it do the work of generating the necessary low level 
code.  This would be way better than forcing every library that wants to do 
something cross-platform but low-level write asm or wacky C stuff or 
include ffcall or another redundant cross-platform low level library (like 
GNU pth, or whatever).  If we got the generator function(s) right, you 
could write a caml library that does wacky stuff and do it in a 
cross-platform way (and if the meta-function was right and well debugged 
you wouldn't have to exhaustively test your new library on every platform 
before releasing it...well, maybe that's a bit optimistic, but you get the 
idea).

The problem is to define the correct meta-function (or set of them), and 
I'm not sure what that would be yet, but I'm pretty sure it's possible.  I 
got the idea from the concept of the "fold" function (homomorphisms)...you 
can use fold to generate tons of other useful functions.  I want the fold 
for platform-specific low-level craziness.  :)

Enough rambling,
Chris


-------------------
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] CamlIDL question

[Christian Stork]

On Wed, Oct 26, 2005 at 01:03:45AM -0400, Igor Pechtchanski wrote:
> On Tue, 25 Oct 2005, Christian Stork wrote:

> > I must be making a stupid mistake but I just don't see it.  I put together a
> > simple example that shows the problem.  Even though the camlidl library is
> > found its symbol "camlidl_free" is not.  How can that be?  Can anybody
> > reproduce the problem or is this some hickup on my system?

> > Here's what make produces (all files attached):
> > [snip]
> > make extest
> > ocamlc -verbose -custom -cclib -lcamlidl -o extest \
> >     ex.o ex_stubs.o ex.cmo extest.cmo

> Shouldn't this be

> ocamlc -verbose -custom -cclib -o extest \
>     ex.o ex_stubs.o ex.cmo extest.cmo -lcamlidl

> instead?  Libraries should follow objects that refer to them...

Thanks, that was it.

-- 
Chris Stork   <>  Support eff.org!  <>   http://www.ics.uci.edu/~cstork/
OpenPGP fingerprint:  B08B 602C C806 C492 D069  021E 41F3 8C8D 50F9 CA2F

Re: [Caml-list] CamlIDL question

[Igor Pechtchanski]

On Tue, 25 Oct 2005, Christian Stork wrote:

> I must be making a stupid mistake but I just don't see it.  I put together a
> simple example that shows the problem.  Even though the camlidl library is
> found its symbol "camlidl_free" is not.  How can that be?  Can anybody
> reproduce the problem or is this some hickup on my system?
>
> Here's what make produces (all files attached):
> [snip]
> make extest
> ocamlc -verbose -custom -cclib -lcamlidl -o extest \
>     ex.o ex_stubs.o ex.cmo extest.cmo

Shouldn't this be

ocamlc -verbose -custom -cclib -o extest \
    ex.o ex_stubs.o ex.cmo extest.cmo -lcamlidl

instead?  Libraries should follow objects that refer to them...
HTH,
	Igor
-- 
				http://cs.nyu.edu/~pechtcha/
      |\      _,,,---,,_		pechtcha@cs.nyu.edu
ZZZzz /,`.-'`'    -.  ;-;;,_		igor@watson.ibm.com
     |,4-  ) )-,_. ,\ (  `'-'		Igor Pechtchanski, Ph.D.
    '---''(_/--'  `-'\_) fL	a.k.a JaguaR-R-R-r-r-r-.-.-.  Meow!

If there's any real truth it's that the entire multidimensional infinity
of the Universe is almost certainly being run by a bunch of maniacs. /DA