Mailing list for all users of the OCaml language and system.
 help / color / mirror / Atom feed
From: Jason Hickey <jyh@cs.caltech.edu>
To: Artem Prisyznuk <tema@sit.kiev.ua>
Cc: caml-bugs@pauillac.inria.fr, caml-list@inria.fr
Subject: Re: [Caml-list] Troubles with marshaled/unmarshaled exception
Date: Thu, 04 Dec 2003 16:14:40 -0800	[thread overview]
Message-ID: <3FCFCDF0.3070606@cs.caltech.edu> (raw)
In-Reply-To: <oprznx3sjeafbmjd@uaapc442>

The short answer is: don't marshal exceptions.

The longer answer is that exception marshaling is nearly as hard as 
function/closure marshaling.  I'm sure the OCaml implementors can give a 
more precise description, but here is my understanding.

First,
    1. A program can have a large number of exceptions.
    2. Exception names are scoped.  The following exceptions
       A.Foo and B.Foo are different.  The same holds for
       compilation units/files that are separately compiled.

       module A = struct exception Foo end
       module B = struct exception Foo end

Just like normal unions, the exceptions are eventually represented as a 
block with a tag and some values, but exception tags are resolved at 
*link* time.

This explains why you can't marshal an exception in one process, send it 
over a channel, unmarshal it in another process, and expect it to make 
any sense at all, because the linker's tag selection is likely to be 
different in the two processes.

Why doesn't it work within the same process?  This is most likely 
because the exception tag is a (pointer to a) normal OCaml value, but 
pattern matching on the exception value uses pointer equality for the 
tag.  When you marshal/unmarshal, you deep-copy the tag, and pointer 
equality no longer works (though normal equality works just fine). 
There are excellent implementation reasons to use pointer equality on 
tags; and faithfulness of marshaling is a low priority.

So, it is likely the marshaler could be changed so that exception 
marshaling would work as you might expect, but it would be hard, and not 
very worthwhile.  Don't expect exception marshaling to work in a 
portable manner...

Jason

Artem Prisyznuk wrote:

> Hello,
> 
> I found strange behavior of pattern matching of
> marshaled/unmarshaled exception.
> 
> Next code describe problem:
> 
>     let e = Failure "test";;
>     let e' = Marshal.from_string (Marshal.to_string e []) 0;;
>     let print_fun exc =
>       match exc with
>         Failure _ -> print_endline "Matching OK"
>       | exc -> print_endline "Matching Fail";;
> 
>     print_fun e;;
>     print_fun e';;
>       Printf.printf "e = e' is %b\n" (e=e');;
> 
> Output:
> 
>     Matching OK
>     Matching Fail
>       e = e' is true
> 
> So second call print missing value.
> 

-- 
Jason Hickey                  http://www.cs.caltech.edu/~jyh
Caltech Computer Science      Tel: 626-395-6568 FAX: 626-792-4257

-------------------
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


  parent reply	other threads:[~2003-12-05  0:14 UTC|newest]

Thread overview: 8+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2003-12-04 15:46 Artem Prisyznuk
2003-12-04 18:11 ` Alex Baretta
2003-12-04 18:51   ` Eric C. Cooper
2003-12-04 19:05     ` Alex Baretta
2003-12-04 19:50       ` Eric C. Cooper
2003-12-04 19:14   ` Artem Prisyznuk
2003-12-05  0:14 ` Jason Hickey [this message]
2003-12-05  7:31   ` Artem Prisyznuk

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=3FCFCDF0.3070606@cs.caltech.edu \
    --to=jyh@cs.caltech.edu \
    --cc=caml-bugs@pauillac.inria.fr \
    --cc=caml-list@inria.fr \
    --cc=tema@sit.kiev.ua \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox