marshalling objects

marshalling objects

[Pietro Abate]

Hi all,

It's not clear to me if it is possible or not to marshall objects ? From
my tests it is not the case (output_value: abstract value (outside
heap).). Am I missing something ? What is common practice to send
objects across the wire ?

thanks,
p

--------- ex.ml

class node = object val d = 1 end ;;

let l =
    try
        let a = (new node) in
        let s = (Marshal.to_string a [])
        in print_endline "Success"; (Marshal.from_string s 0 : node)
    with Invalid_argument s -> print_endline ("Failed "^s) ; (new node)
;;

let l =
    try
        let a = (new node) in
        let s = (Marshal.to_string a [Marshal.Closures])
        in print_endline "Success"; (Marshal.from_string s 0 : node)
    with Invalid_argument s -> print_endline ("Failed "^s) ; (new node)
;;

let l =
    try
        let a = (new node) in
        let s = (Marshal.to_string a [Marshal.No_sharing])
        in print_endline "Success"; (Marshal.from_string s 0 : node)
    with Invalid_argument s -> print_endline ("Failed "^s) ; (new node)
;;

let l =
    try
        let a = (new node) in
        let s = (Marshal.to_string a [Marshal.No_sharing;Marshal.Closures])
        in print_endline "Success"; (Marshal.from_string s 0 : node)
    with Invalid_argument s -> print_endline ("Failed "^s) ; (new node)
;;


-- 
++ Blog: http://blog.rsise.anu.edu.au/?q=pietro
++ 
++ "All great truths begin as blasphemies." -George Bernard Shaw
++ Please avoid sending me Word or PowerPoint attachments.
   See http://www.fsf.org/philosophy/no-word-attachments.html

Re: [Caml-list] marshalling objects

[Jacques Garrigue]

From: Pietro Abate <Pietro.Abate@anu.edu.au>

> It's not clear to me if it is possible or not to marshall objects ? From
> my tests it is not the case (output_value: abstract value (outside
> heap).). Am I missing something ? What is common practice to send
> objects across the wire ?

It is possible, but there was a bug in 3.08.
In 3.09, you can marshal an object with
   Marshal.to_string obj [Marshal.Closures]
Note that this means that the method table is really sent, as an array
of closures, which is possibly very innefficient, and also means that
the program on the other side must be the same one.

Jacques Garrigue

Re: [Caml-list] marshalling objects

[sejourne_kevin]

Jacques Garrigue a écrit :
> Note that this means that the method table is really sent, as an array
> of closures, which is possibly very innefficient, and also means that
> the program on the other side must be the same one.

Exactly the same one or just all classes ?
I think this could be in the manual of module Marshal.


Kévin.

	

	
		
___________________________________________________________________________ 
Appel audio GRATUIT partout dans le monde avec le nouveau Yahoo! Messenger 
Téléchargez cette version sur http://fr.messenger.yahoo.com

Re: [Caml-list] marshalling objects

[Jacques Garrigue]

From: sejourne_kevin <sejourne_kevin@yahoo.fr>
> Jacques Garrigue a écrit :
> > Note that this means that the method table is really sent, as an array
> > of closures, which is possibly very innefficient, and also means that
> > the program on the other side must be the same one.
> 
> Exactly the same one or just all classes ?
> I think this could be in the manual of module Marshal.

Actually it's just a consequence of the internal representation of
classes, which is described in the C interface section.
That is, each object contains an array with all its methods, which are
function closures.
This is the reason you need the flag for closures.
And the condition with closures is quite explicit: this must be
exactly the same program. I also expect that, for native code, it must
be the same architecture.

Jacques Garrigue

marshalling objects

[Jacques Le Normand]

[-- Attachment #1: Type: text/plain, Size: 217 bytes --]

hello caml list,
if I marshal an object of class c and I write the result in a file foo.txt
and, later on, I add methods to c, can I read the object in foo.txt and call
the new methods on the object?
cheers
--Jacques

[-- Attachment #2: Type: text/html, Size: 231 bytes --]