[Caml-list] pretty printers and format and matrices

[Caml-list] pretty printers and format and matrices

[Chris Hecker]

Format has come up on the list a couple times, so I figured I'd ask a question I gave up on trying to solve a while back.

I'd like to write a printer for the top level that will print out matrices as rectangular blocks of numbers, not lists of numbers.  I couldn't figure out whether this was possible with install_printer in a clean way.  The obvious things didn't work quite right.

I finally gave up and ended up with this (open Bigarray first):

let print2d a =
  for i = 0 to Array2.dim1 a - 1 do
    for j = 0 to Array2.dim2 a - 1 do
      Format.printf "%f\t" a.{i,j}
    done;
    Format.printf "\n";
  done

But this doesn't play nice with other printers like it should.  I tried a few different attemps at putting the boxes in, but none worked.  I also read the FAQ, but couldn't quite apply its info to this problem.  Is there a "right" way to do this, or is the Format module not set up for vertical printing?

Chris

-------------------
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] pretty printers and format and matrices

[Markus Mottl]

On Tue, 12 Jun 2001, Chris Hecker wrote:
> Format has come up on the list a couple times, so I figured I'd ask
> a question I gave up on trying to solve a while back.

Once one gets used to it, it's really fun so don't give up...

> I finally gave up and ended up with this (open Bigarray first):
> 
> let print2d a =
>   for i = 0 to Array2.dim1 a - 1 do
>     for j = 0 to Array2.dim2 a - 1 do
>       Format.printf "%f\t" a.{i,j}
>     done;
>     Format.printf "\n";
>   done

The reason is obviously the use of "\n": this is a "real" newline, which
disregards indentation levels. The pretty printer will still assume that
it hasn't broken lines if you use this, which makes further output look
really strange.

> But this doesn't play nice with other printers like it should.  I tried
> a few different attemps at putting the boxes in, but none worked.
> I also read the FAQ, but couldn't quite apply its info to this problem.
> Is there a "right" way to do this, or is the Format module not set up
> for vertical printing?

You'll have to use "@\n" instead or also "Format.force_newline ()"
if you want to force a newline without confusing indentation levels.

The Format-module is very powerful! In most cases it will suffice if
you use hov-boxes (the default with "@[") + indentation annotations as
required. Try this for instance:

---------------------------------------------------------------------------
open Bigarray
open Format

let print2d ppf a =
  fprintf ppf "@[<2>[|";
  for i = 0 to Array2.dim1 a - 1 do
    pp_force_newline ppf ();
    for j = 0 to Array2.dim2 a - 1 do
      fprintf ppf "%f;\t" a.{i,j}
    done
  done;
  fprintf ppf "@]@\n|]"

let _ =
  let ar = Array2.of_array float32 c_layout [|[|1.; 2.|]; [|3.; 4.|]|] in
  printf "%a@\n" print2d ar
---------------------------------------------------------------------------

Regards,
Markus Mottl

-- 
Markus Mottl, mottl@miss.wu-wien.ac.at, http://miss.wu-wien.ac.at/~mottl
-------------------
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] pretty printers and format and matrices

[Chris Hecker]

>Once one gets used to it, it's really fun so don't give up...

I had the choice of "try to get the formatting working for no other reason than curiosity" or "write more real code on my game", so the latter edged it out.  :)

> let print2d ppf a =

This works better, thanks!  I swear I tried something exactly like that...

There's a tiny bit of weirdness with the tabs after the first column, but it's way better than what I had.  I definitely tried the @\n, but I admit I was randomly typing at that point.  The docs could stand to be a bit clearer on newlines versus breaks.  Perhaps your print2d would make a good complementary tutorial to the expression printer in the FAQ...

Thanks,
Chris

-------------------
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] pretty printers and format and matrices

[Hendrik Tews]

Hi, 

I have another question about the format module:

How can I set max_boxes to infinity?

We use the format module to output source code of a programming
language. For that one needs a way to advice the format module to
print everything, regardless of how deep boxes are nested. At the
moment we use (set_max_boxes 1000), and when we run into problems
we increase the number.


Bye,

Hendrik
-------------------
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] pretty printers and format and matrices

[Pierre Weis]

> Hi, 
> 
> I have another question about the format module:
> 
> How can I set max_boxes to infinity?
> 
> We use the format module to output source code of a programming
> language. For that one needs a way to advice the format module to
> print everything, regardless of how deep boxes are nested. At the
> moment we use (set_max_boxes 1000), and when we run into problems
> we increase the number.
> 
> 
> Bye,
> 
> Hendrik
> -------------------
> 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

Having to open more than 1000 boxes to print any material seems
incredible to me, since I used to pretty print an entire Caml compiler
within 35 boxes.

Anyway, I'm afraid there is no notion of infinity in Format. Neither
in the Caml integer values. Hence no solution to your problem.

The best you can do is to use a rough approximation, as in:

set_max_boxes max_int;;

On a regular machine this allows to simultaneously open no more than
4611686018427387903 boxes, which could be large enough to pretty print
a huge source chunk of your programming language source programs. On
the other hand, on my Ipaq handheld this limit is much smaller than
that, namely 1073741823 which may be really short for you.

If you need more than max_int simultaneously open boxes, you should
rewrite the Format module using the bignumbers facility of
Caml. However, I should admit that you will still be facing another
stupid limit (no more than 2^30 words in a big integer, which means
approximately 9 * 2 ^ 30 decimal digits on a little machine, once more
far far away from infinity). If you really need infinity, I am not aware
of any practical solution in Objective Caml.

Sorry for this unsatisfactory answer, due to stupid limitations in the
language.

Pierre Weis

INRIA, Projet Cristal, Pierre.Weis@inria.fr, http://pauillac.inria.fr/~weis/


-------------------
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] pretty printers and format and matrices

[Patrick M Doane]

On Thu, 14 Jun 2001, Pierre Weis wrote:

> Having to open more than 1000 boxes to print any material seems
> incredible to me, since I used to pretty print an entire Caml compiler
> within 35 boxes.
> 
> Anyway, I'm afraid there is no notion of infinity in Format. Neither
> in the Caml integer values. Hence no solution to your problem.

Would it be reasonable to increase the default value from 35?  Are there
any performance problems with setting this value to max_int in the
pp_make_formatter routine?

Patrick

-------------------
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] pretty printers and format and matrices

[Pierre Weis]

> On Thu, 14 Jun 2001, Pierre Weis wrote:
> 
> > Having to open more than 1000 boxes to print any material seems
> > incredible to me, since I used to pretty print an entire Caml compiler
> > within 35 boxes.
> > 
> > Anyway, I'm afraid there is no notion of infinity in Format. Neither
> > in the Caml integer values. Hence no solution to your problem.
> 
> Would it be reasonable to increase the default value from 35?  Are there
> any performance problems with setting this value to max_int in the
> pp_make_formatter routine?
> 
> Patrick
> 
> -------------------
> 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

Yes, setting the default value to the maximum possible value is
conceivable. However, this (reasonable) limit is there as a kind of
alert to tell the user that the number of boxes the programs opens is
now considered large (hence you may consider having a look at the
printing program to see if it works properly). This is the performance
problem you may have (not in the library but in your printing
routines). However if this is generally agreed, we can do it ...

Best regards,

Pierre Weis

INRIA, Projet Cristal, Pierre.Weis@inria.fr, http://pauillac.inria.fr/~weis/


-------------------
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] pretty printers and format and matrices

[Frank Atanassow]

Pierre Weis wrote (on 14-06-01 21:37 +0200):
> On a regular machine this allows to simultaneously open no more than
> 4611686018427387903 boxes, which could be large enough to pretty print
> a huge source chunk of your programming language source programs. On
> the other hand, on my Ipaq handheld this limit is much smaller than
> that, namely 1073741823 which may be really short for you.
> 
> If you need more than max_int simultaneously open boxes, you should
> rewrite the Format module using the bignumbers facility of
> Caml. However, I should admit that you will still be facing another
> stupid limit (no more than 2^30 words in a big integer, which means
> approximately 9 * 2 ^ 30 decimal digits on a little machine, once more
> far far away from infinity). If you really need infinity, I am not aware
> of any practical solution in Objective Caml.
> 
> Sorry for this unsatisfactory answer, due to stupid limitations in the
> language.

Obviously the solution is to rewrite Format so that the box numbers are
encoded as Church numerals. Then you can easily represent infinity:

  let zero = fun z s -> z;;
  let succ n = fun z s -> let s' = s () in s' (n z s);;
  (* unfortunately, eta-expansion needed because of value restriction *)
  let one = fun z s -> succ zero z s;;
  let two = fun z s -> succ one z s;;
  let three = fun z s -> succ two z s;;
  ...
  let rec infty = fun z s -> let s' = s () in s' (infty z s);;

  let eq m n = m (fun () 
  let add m n = m n (fun () -> succ);;
  let print_num n = n () (fun () -> print_string "1"; fun _ -> ());;

  # print_num one;;
  1- : unit = ()
  # print_num (add two three);;
  11111- : unit = ()
  # print_num infty;;
  111111111111111111111111111111111111111111111111111111111111111111111111
  111111111111111111111111111111111111111111111111111111111111111111111111
  111111111111111111111111111111111111111111111111111111111111111111111111
  11111111111111111111Interrupted.
  # print_num (succ infty);;
  111111111111111111111111111111111111111111111111111111111111111111111111
  111111111111111111111111111111111111111111111111111111111111111111111111
  111111111111111111111111111111111111111111111111111111111111111111111111
  111111111111111111111Interrupted.

P.S.: When I started writing this, I thought Pierre's post was in jest. Now
I'm not so sure anymore...!

-- 
Frank Atanassow, Information & Computing Sciences, Utrecht University
Padualaan 14, PO Box 80.089, 3508 TB Utrecht, Netherlands
Tel +31 (030) 253-3261 Fax +31 (030) 251-379
-------------------
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] pretty printers and format and matrices

[Pierre Weis]

> P.S.: When I started writing this, I thought Pierre's post was in jest. Now
> I'm not so sure anymore...!

Nor am I!

:)

Pierre Weis

INRIA, Projet Cristal, Pierre.Weis@inria.fr, http://pauillac.inria.fr/~weis/


-------------------
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] pretty printers and format and matrices

[Hendrik Tews]

Hi,

Pierre Weis writes:
   Date: Thu, 14 Jun 2001 21:37:44 +0200 (MET DST)
   Subject: Re: [Caml-list] pretty printers and format and matrices
   
   However, I should admit that you will still be facing another
   stupid limit (no more than 2^30 words in a big integer, which means
   approximately 9 * 2 ^ 30 decimal digits on a little machine, once more
   far far away from infinity). 

OK, I see you are in sarcastic mood ... I am actually negotiating
with John Skaller the foundation of the "Got bitten by Pierre
Club", anybody else who wants to participate?
;-)

   Sorry for this unsatisfactory answer, due to stupid limitations in the
   language.
   
Then I would like to suggest to add a new feature to a future
version of the format module: "do_never_print_ellipsis". Perhaps 
(set_max_boxes -1) could be interpreded this way.

Problems like "The most interesting problem repaired by the
Judges was that one entry nested Objective Caml's Format boxes
too deeply ....", quoted from [1], would then belong to the past.


BTW:

   set_max_boxes max_int;;
   
   On a regular machine this allows to simultaneously open no more than
   4611686018427387903 boxes, which could be large enough to pretty print

I have 

# max_int;;
- : int = 1073741823

so what's a regular machine??


Thanks anyway for the unsatisfactory answer,

Hendrik


[1] The 1999 ICFP Programming Contest. ACM SIGPLAN Notices,
35(3), pp. 73-83, March 2000. Available at
http://www.eecs.harvard.edu/~nr/pubs/icfp99.ps
-------------------
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] pretty printers and format and matrices

[Pierre Weis]

Hi,
 
> Pierre Weis writes:
>    Date: Thu, 14 Jun 2001 21:37:44 +0200 (MET DST)
>    Subject: Re: [Caml-list] pretty printers and format and matrices
>    
>    However, I should admit that you will still be facing another
>    stupid limit (no more than 2^30 words in a big integer, which means
>    approximately 9 * 2 ^ 30 decimal digits on a little machine, once more
>    far far away from infinity). 
> 
> OK, I see you are in sarcastic mood ... I am actually negotiating
> with John Skaller the foundation of the "Got bitten by Pierre
> Club", anybody else who wants to participate?
> ;-)

Sorry for that I was just a bit tired by a more than 5 hours long
meeting. I was also taking it for granting that everybody would
understand that I was joking: could you imagine opening so much boxes
that it could overflow a limit which is as large as a number that has
``9 * 2 ^ 30 decimal digits'' ? Could you imagine anything real (such
that a computation or something that you can count) that can overflow
such a huge number (much much bigger than even a google) ?

Also, I don't know if we have to permanently be calm and solemn when
posting to this mailing list. Well, I guess it's the best way given
the medium ...

>    Sorry for this unsatisfactory answer, due to stupid limitations in the
>    language.
>    
> Then I would like to suggest to add a new feature to a future
> version of the format module: "do_never_print_ellipsis". Perhaps 
> (set_max_boxes -1) could be interpreded this way.

This way of interpreting negative numbers seems a bit strange to
me. Also this would add a lot of particular cases, just for people who
need more than 1073741823 simultaneously open boxes (just have a look
to this number, it is really huge indeed).

> Problems like "The most interesting problem repaired by the
> Judges was that one entry nested Objective Caml's Format boxes
> too deeply ....", quoted from [1], would then belong to the past.

OK, you win. Then I would prefer to set pp_max_boxes to max_int at
creation time of the ``formatter''. This is simpler and more
logical...

> BTW:
> 
>    set_max_boxes max_int;;
>    
>    On a regular machine this allows to simultaneously open no more than
>    4611686018427387903 boxes, which could be large enough to pretty print
> 
> I have 
> 
> # max_int;;
> - : int = 1073741823
> 
> so what's a regular machine??

Sorry for that also, you need some context to understand that one: a
``regular'' machine is a kind of nickname we used here to name our 64
bits alpha machines, I suppose that this is a way to insist on the
fact that there exist other processors than Intel's ...

> Thanks anyway for the unsatisfactory answer,

You're very welcome.

> Hendrik
> 
> 
> [1] The 1999 ICFP Programming Contest. ACM SIGPLAN Notices,
> 35(3), pp. 73-83, March 2000. Available at
> http://www.eecs.harvard.edu/~nr/pubs/icfp99.ps

Regards,

Pierre Weis

INRIA, Projet Cristal, Pierre.Weis@inria.fr, http://pauillac.inria.fr/~weis/


-------------------
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] pretty printers and format and matrices

[Pierre Weis]

> 
> >Once one gets used to it, it's really fun so don't give up...
> 
> I had the choice of "try to get the formatting working for no other
> reason than curiosity" or "write more real code on my game", so the
> latter edged it out.  :) 
> 
> > let print2d ppf a =
> 
> This works better, thanks!  I swear I tried something exactly like that...
> 
> There's a tiny bit of weirdness with the tabs after the first
> column, but it's way better than what I had.  I definitely tried the
> @\n, but I admit I was randomly typing at that point.  The docs
> could stand to be a bit clearer on newlines versus breaks.  Perhaps
> your print2d would make a good complementary tutorial to the
> expression printer in the FAQ... 
> 
> Thanks,
> Chris

Have you read the special paragraph devoted to Format in the FAQ ?

http://pauillac.inria.fr/caml/FAQ/format-eng.html

It says somewhere:

Practice
When writing a pretty-printing routine, follow these simple rules: 

[...]
4.Don't try to force line breaking, let the pretty-printer do it for
you: that's its only job. 
5.End your main program by a print_newline () call, that flushes the
pretty-printer tables (hence the ouput). (Note that the toplevel loop
of the interactive system does it as well, just before a new input.) 

Best regards,

Pierre Weis

INRIA, Projet Cristal, Pierre.Weis@inria.fr, http://pauillac.inria.fr/~weis/


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