[Caml-list] error messages to stdout?

[Caml-list] error messages to stdout?

[Doug Bagley]

I've been learning how to use Ocaml for writing scripts, and I've
encountered a small problem.

By "scripts", I mean they have the canonical custom toplevel in the
interpreter line.  They also dynamically load libs.

I had updated one lib and not another that depended on it, so I got a
message like so, to stdout:

 File /usr/local/lib/ocaml/contrib/fs_lib.cma is not up-to-date with respect to
 interface Pcre

The problem was that I piped my ocaml script's output to a file that
was supposed to just contain data, and then I graphed it, so the error
message showed up in a graph, which was kind of funny but not what I
expected. I use pipes and filters a lot, so this could really put a
wrench in my plans.

Anyway, it would be like, cool, if I could get those errors going to
stderr instead, or be enlightened as to the error of my ways. Thanks.

cheers,
doug
-------------------
Bug reports: http://caml.inria.fr/bin/caml-bugs  FAQ: http://caml.inria.fr/FAQ/
To unsubscribe, mail caml-list-request@inria.fr  Archives: http://caml.inria.fr

Re: [Caml-list] error messages to stdout?

[Warp]

>  File /usr/local/lib/ocaml/contrib/fs_lib.cma is not up-to-date with
respect to
>  interface Pcre
>
> The problem was that I piped my ocaml script's output to a file that
> was supposed to just contain data, and then I graphed it, so the error
> message showed up in a graph, which was kind of funny but not what I
> expected. I use pipes and filters a lot, so this could really put a
> wrench in my plans.
>
> Anyway, it would be like, cool, if I could get those errors going to
> stderr instead, or be enlightened as to the error of my ways. Thanks.

If this message comes from RunTime ( when Dynamic Linking ) then that might
be an exception throwed by the dynlink. Then you can catch it and print it
to stderr ( using Printexc.to_string )

-------------------
Bug reports: http://caml.inria.fr/bin/caml-bugs  FAQ: http://caml.inria.fr/FAQ/
To unsubscribe, mail caml-list-request@inria.fr  Archives: http://caml.inria.fr

Re: [Caml-list] error messages to stdout?

[Doug Bagley]

Warp wrote:
> If this message comes from RunTime ( when Dynamic Linking ) then
> that might be an exception throwed by the dynlink. Then you can
> catch it and print it to stderr ( using Printexc.to_string )

Well, I may be wrong, but this message is just printed to stdout by
toplevel/topdirs.ml ... it is not the result of an exception. And, I'm
no expert, but I believe "exceptions" already go to stderr.

The philosophy of printing all errors and diagnostics to a separate
channel from program data is, well, maybe the way to go? Just my .02
Euros.

cheers,
doug
-------------------
Bug reports: http://caml.inria.fr/bin/caml-bugs  FAQ: http://caml.inria.fr/FAQ/
To unsubscribe, mail caml-list-request@inria.fr  Archives: http://caml.inria.fr

Re: [Caml-list] error messages to stdout?

[Remi VANICAT]

Doug Bagley <doug@bagley.org> writes:

> I've been learning how to use Ocaml for writing scripts, and I've
> encountered a small problem.
> 
> By "scripts", I mean they have the canonical custom toplevel in the
> interpreter line.  They also dynamically load libs.
> 
> I had updated one lib and not another that depended on it, so I got a
> message like so, to stdout:
> 
>  File /usr/local/lib/ocaml/contrib/fs_lib.cma is not up-to-date with respect to
>  interface Pcre
> 
> The problem was that I piped my ocaml script's output to a file that
> was supposed to just contain data, and then I graphed it, so the error
> message showed up in a graph, which was kind of funny but not what I
> expected. I use pipes and filters a lot, so this could really put a
> wrench in my plans.
> 
> Anyway, it would be like, cool, if I could get those errors going to
> stderr instead, or be enlightened as to the error of my
> ways. Thanks.

you should compile your program to (at least) bytecode code. It will
be faster, and those kind of error will only exist at compile time,
not at execution time...
-------------------
Bug reports: http://caml.inria.fr/bin/caml-bugs  FAQ: http://caml.inria.fr/FAQ/
To unsubscribe, mail caml-list-request@inria.fr  Archives: http://caml.inria.fr

Re: [Caml-list] error messages to stdout?

[Doug Bagley]

Remi VANICAT wrote:
> you should compile your program to (at least) bytecode code. It will
> be faster, and those kind of error will only exist at compile time,
> not at execution time...

Well, I certainly could do that, but I was hoping to use Ocaml to
replace my bash/awk/sed/perl scripts. For me, the convenience of
having one single source file, and no compilation step, is one of the
attractions of "scripts". As far as I can tell, the extra load time
due to compilation of the script is negligible, and I assume the
run-time is equivalent to a pre-compiled byte-code executable? I have
many scripts that do filtering, so having a clean output stream is
important to me.  It seems to me that if the ability to use a custom
toplevel as a script interpreter exists, then it would be very nice if
it printed all diagnostics to stderr.

Anyway, I have a workaround, I wrote a "trampoline" program that can
be used in place of the toplevel interpreter and does the compilation
transparently for lazy persons such as myself.

It essentially turns any compiled language into a scripting language.
If anyone is interested in it ... I could clean it up for public
consumption.

cheers,
doug
-------------------
Bug reports: http://caml.inria.fr/bin/caml-bugs  FAQ: http://caml.inria.fr/FAQ/
To unsubscribe, mail caml-list-request@inria.fr  Archives: http://caml.inria.fr

Re: [Caml-list] error messages to stdout?

[John Malecki]

Doug Bagley wrote (2002-01-16T14:09:48-0600):
 > 
 > Anyway, I have a workaround, I wrote a "trampoline" program that can
 > be used in place of the toplevel interpreter and does the compilation
 > transparently for lazy persons such as myself.
 > 
 > It essentially turns any compiled language into a scripting language.
 > If anyone is interested in it ... I could clean it up for public
 > consumption.

Hi Doug,

I'd be interested in the trampoline program.

At the moment we are operating in a homogeneous computing environment.
Sooner or later we will add some a different computer architecture to
our pool and I was thinking about how to take advantage of compiling
byte-code once and using it on all platforms.  This has the advantage
that a developer can compile once, on any one platform, and make that
program immediately available on all other machine architectures.

I'm not sure of the best way to do this.  I was thinking of using the
bash MACHTYPE environment variable and then install the byte-code
program in /usr/local/bin but have the 1st line of that file say

   #! /usr/local/bin/ocamlrun.$MACHTYPE

Unfortunately the loader does not perform environment variable
expansion.  I tried other tricks but I couldn't find a simple "1
liner".  I'm still looking for something simple.  If anyone has any
ideas I'd like to hear them.

(Yes, there are probably automount tricks I could use but I'm looking
for a solution that will work for people without administrator
privileges.)


-cheers
-------------------
Bug reports: http://caml.inria.fr/bin/caml-bugs  FAQ: http://caml.inria.fr/FAQ/
To unsubscribe, mail caml-list-request@inria.fr  Archives: http://caml.inria.fr

Re: [Caml-list] error messages to stdout?

[Doug Bagley]

John Malecki wrote:
> Doug Bagley wrote (2002-01-16T14:09:48-0600):
>  > 
>  > Anyway, I have a workaround, I wrote a "trampoline" program that can
>  > be used in place of the toplevel interpreter and does the compilation
>  > transparently for lazy persons such as myself.
>  > 
>  > It essentially turns any compiled language into a scripting language.
>  > If anyone is interested in it ... I could clean it up for public
>  > consumption.
> 
> Hi Doug,
> 
> I'd be interested in the trampoline program.

Well, it's pre-release, and uses some pre-requesite libs of mine that
are also pre-release :-)  And I haven't had anyone else test out the
build process yet, and so on and so forth, but if you are adventurous,
you are welcome to take a look and give me feedback, as long as you
realize it is a very raw work-in-progress :)

It is currently here:

 http://www.bagley.org/~doug/ocaml/scriptomatic/

It basically hands off the build work to make, and I have a very
simple makefile included whichs needs a lot of polishing, but the
basic idea seems sound, and I can write "scripts" that compile to
byte or optimized Ocaml, or even write "scripts" in C :) The built
executables are cached, and only rebuilt if the original script
changes.

> At the moment we are operating in a homogeneous computing environment.
> Sooner or later we will add some a different computer architecture to
> our pool and I was thinking about how to take advantage of compiling
> byte-code once and using it on all platforms.  This has the advantage
> that a developer can compile once, on any one platform, and make that
> program immediately available on all other machine architectures.
> 
> I'm not sure of the best way to do this.  I was thinking of using the
> bash MACHTYPE environment variable and then install the byte-code
> program in /usr/local/bin but have the 1st line of that file say
> 
>    #! /usr/local/bin/ocamlrun.$MACHTYPE
> 
> Unfortunately the loader does not perform environment variable
> expansion.  I tried other tricks but I couldn't find a simple "1
> liner".  I'm still looking for something simple.  If anyone has any
> ideas I'd like to hear them.

I think that simply doing what you want above is a little easier than
what I have implemented, and you may wish to use the simplest solution.
I've appended an idea for a replacement interpreter that is an "almost
one-liner" in C below. It switches the toplevel interpreter based on
the machine type environement variable. Maybe it will give you an idea.

(Note that on Linux, the program execed by the trampoline may not see
itself in argv[0], I sent a note to caml-bugs about an old fix (557),
which I think is related to the problem).

cheers,
doug
-- 
trampo.c:
/* gcc trampo.c -o trampo */
#include <stdio.h>
main (int argc, char **argv) {
    char exe[256];
    if (snprintf(exe, 256, "ocamlrun.%s", getenv("MACHTYPE")) > -1)
	execvp(exe, argv);
    else fprintf(stderr, "ouch\n");
}
/* end */

int the shell:
export MACHTYPE=MACLISA  # or whatever

tramptest:
#! ./trampo
let _ = print_endline "Hello World"
-------------------
Bug reports: http://caml.inria.fr/bin/caml-bugs  FAQ: http://caml.inria.fr/FAQ/
To unsubscribe, mail caml-list-request@inria.fr  Archives: http://caml.inria.fr

Re: [Caml-list] error messages to stdout?

[Stefano Zacchiroli]

On Thu, Jan 17, 2002 at 10:33:06AM -0800, John Malecki wrote:
> At the moment we are operating in a homogeneous computing environment.
> Sooner or later we will add some a different computer architecture to
> our pool and I was thinking about how to take advantage of compiling
> byte-code once and using it on all platforms.  This has the advantage
> that a developer can compile once, on any one platform, and make that
> program immediately available on all other machine architectures.
> 
> I'm not sure of the best way to do this.  I was thinking of using the
> bash MACHTYPE environment variable and then install the byte-code
> program in /usr/local/bin but have the 1st line of that file say
> 
>    #! /usr/local/bin/ocamlrun.$MACHTYPE

Probably I miss the point but if your machines have some local disk data
you can install ocamlrun executable locally on each machine but using
the same path. In such a way using "#! /usr/local/bin/ocamlrun" will
work on each architecture.

But, obviously, if all of your filesystems are shared this trick doesn't
work.

Cheers.

-- 
Stefano "Zack" Zacchiroli <zack@cs.unibo.it> ICQ# 33538863
Home Page: http://www.cs.unibo.it/~zacchiro
Undergraduate student of Computer Science @ University of Bologna, Italy
                 - Information wants to be Open -
-------------------
Bug reports: http://caml.inria.fr/bin/caml-bugs  FAQ: http://caml.inria.fr/FAQ/
To unsubscribe, mail caml-list-request@inria.fr  Archives: http://caml.inria.fr

Re: [Caml-list] error messages to stdout?

[Florian Douetteau]

>
> I'd be interested in the trampoline program.
>
> At the moment we are operating in a homogeneous computing environment.
> Sooner or later we will add some a different computer architecture to
> our pool and I was thinking about how to take advantage of compiling
> byte-code once and using it on all platforms.  This has the advantage
> that a developer can compile once, on any one platform, and make that
> program immediately available on all other machine architectures.
>
> I'm not sure of the best way to do this.  I was thinking of using the
> bash MACHTYPE environment variable and then install the byte-code
> program in /usr/local/bin but have the 1st line of that file say
>
>    #! /usr/local/bin/ocamlrun.$MACHTYPE
>

An usual trick  is to make the bytecode file
start with

#! /path_to_my_mach_type_dispatcher /usr/local/bin/ocamlrun

Executing the file will make the system execute:
"my_mach_type_dispatcher /usr/local/bin/ocamlrun bytecode-file-name"

my_mach_type_dispatcher can be a script as simple as:
#! /bin/sh
$1.$MACHTYPE $2

Another trick is to make each /usr/local/bin/ocamlrun.$MACHTYPE
be a symbol link to a script:
$0.$MACHTYPE $@

--
Florian Douetteau
-------------------
Bug reports: http://caml.inria.fr/bin/caml-bugs  FAQ: http://caml.inria.fr/FAQ/
To unsubscribe, mail caml-list-request@inria.fr  Archives: http://caml.inria.fr