[Caml-list] duplicate record labels and modules

[Caml-list] duplicate record labels and modules

[briand]

Consider the following :

let a = {Foo.n1=1; Foo.n2=2; Foo.a=(- 1.0)} in
let b = {Bar.n1=4; Bar.n2=(- 5); Bar.b=77.} in
  Printf.printf "%d\n" a.Foo.n1;
  Printf.printf "%d\n" b.Bar.n1;

That's too verbose, and I can't use open or I get the dreaded mixing
labels problem.  So how about:

let (a:Foo.t) =  {n1=1; n2=2; a=(- 1.0)} in

Unbound record field label n1

After careful review of the grammar I discovered that is incorrect,
although it seems like it should work.

A little more perusing and I found:

let-binding ::= pattern = expr 
                | value-name { parameter } [: typexpr] = expr

and that makes sense.  I'm simply explictly annotating the type of a,
right ?

So I tried:

let a:Foo.t =  {n1=1; n2=2; a=(- 1.0)} in

And got the same error, i.e. unbound field n1....

I'm doing a lot of work so I can be lazy :-)


Thanks

Brian

-------------------
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] duplicate record labels and modules

[Ville-Pertti Keinonen]

On May 11, 2004, at 8:13 AM, briand@aracnet.com wrote:

> Consider the following :
>
> let a = {Foo.n1=1; Foo.n2=2; Foo.a=(- 1.0)} in
> let b = {Bar.n1=4; Bar.n2=(- 5); Bar.b=77.} in
>   Printf.printf "%d\n" a.Foo.n1;
>   Printf.printf "%d\n" b.Bar.n1;

If you want to make it slightly less verbose, you can say

module F = Foo
module B = Bar

Or you could use "let module ... in".

Alternately, assuming the modules are your own, you can name your 
record fields so that opening both modules doesn't cause conflicts 
(e.g. using a short prefix).

> let (a:Foo.t) =  {n1=1; n2=2; a=(- 1.0)} in
>
> Unbound record field label n1
>
> After careful review of the grammar I discovered that is incorrect,
> although it seems like it should work.

Nothing about "a" affects the meaning of the expression on the right 
hand side of the "=".

Even if it did, the visibility of identifiers is determined only by 
lexical scope, typing has no effect, which is why even if you have 
"a:Foo.t" in scope, you can't access "a.n1".

-------------------
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] duplicate record labels and modules

[briand]

>>>>> "Ville-Pertti" == Ville-Pertti Keinonen <will@exomi.com> writes:

  Ville-Pertti> On May 11, 2004, at 8:13 AM, briand@aracnet.com wrote:

  >> Consider the following :
  >> 
  >> let a = {Foo.n1=1; Foo.n2=2; Foo.a=(- 1.0)} in
  >> let b = {Bar.n1=4; Bar.n2=(- 5); Bar.b=77.} in
  >> Printf.printf "%d\n" a.Foo.n1;
  >> Printf.printf "%d\n" b.Bar.n1;

  Ville-Pertti> If you want to make it slightly less verbose, you can say

  Ville-Pertti> module F = Foo
  Ville-Pertti> module B = Bar

  Ville-Pertti> Or you could use "let module ... in".

That's useful.  I need to read the chapter on language extensions more
closely.

  Ville-Pertti> Alternately, assuming the modules are your own, you
  Ville-Pertti> can name your record fields so that opening both
  Ville-Pertti> modules doesn't cause conflicts (e.g. using a short
  Ville-Pertti> prefix).

That's very kludgy, isn't it ?  I'm assuming the right way is to use
modules, which is what started this whole exercise for me.


  >> let (a:Foo.t) =  {n1=1; n2=2; a=(- 1.0)} in
  >> 
  >> Unbound record field label n1
  >> 
  >> After careful review of the grammar I discovered that is incorrect,
  >> although it seems like it should work.

  Ville-Pertti> Nothing about "a" affects the meaning of the
  Ville-Pertti> expression on the right hand side of the "=".

  Ville-Pertti> Even if it did, the visibility of identifiers is
  Ville-Pertti> determined only by lexical scope, typing has no
  Ville-Pertti> effect, which is why even if you have "a:Foo.t" in
  Ville-Pertti> scope, you can't access "a.n1".

Oh yes.  That makes makes sense.

Thanks for your help.

Brian

-------------------
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] duplicate record labels and modules

[brogoff]

On Tue, 11 May 2004 briand@aracnet.com wrote:
> That's useful.  I need to read the chapter on language extensions more
> closely.

Also, here's the requisite RTFM (I mean that F in the nicest possible way ;)

	http://caml.inria.fr/FAQ/FAQ_EXPERT-eng.html#labels_surcharge

>   Ville-Pertti> Alternately, assuming the modules are your own, you
>   Ville-Pertti> can name your record fields so that opening both
>   Ville-Pertti> modules doesn't cause conflicts (e.g. using a short
>   Ville-Pertti> prefix).
>
> That's very kludgy, isn't it ?

IMO, it's inconvenient, and a drawback to the language, but not a kludge, or
even a mistake. I reserve that for the undefined order of evaluation :-).
Oh well, you didn't expect paradise, did you?

> I'm assuming the right way is to use modules, which is what started this
> whole exercise for me.

If you must share field names in the same module, your only alternative is to
use the OO part of Ocaml. That's unfortunate, because I think objects are a bit
heavyweight for things like points and segments in a computational geometry
program. OTOH, Ocaml's object system is quite powerful and principled (and
sometimes complex and unintuitive!) compared to lesser languages like C++
and Java. I often think a class system like CLOS or Dylan's would be more to my
liking, but hard to reconcile with ststic typing and modules.

-- Brian

-------------------
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] duplicate record labels and modules

[briand]

>>>>> "brogoff" == brogoff  <brogoff@speakeasy.net> writes:

  brogoff> On Tue, 11 May 2004 briand@aracnet.com wrote:
  >> That's useful.  I need to read the chapter on language extensions
  >> more closely.

  brogoff> Also, here's the requisite RTFM (I mean that F in the
  brogoff> nicest possible way ;)

  brogoff> 	http://caml.inria.fr/FAQ/FAQ_EXPERT-eng.html#labels_surcharge

I Just found it.
It should have occurred to me that this subject would be an FAQ.

However, my original question was more about the syntax of making
using modules to separate the field name less typing-ful :-)

  Ville-Pertti> Alternately, assuming the modules are your own, you
  Ville-Pertti> can name your record fields so that opening both
  Ville-Pertti> modules doesn't cause conflicts (e.g. using a short
  Ville-Pertti> prefix).
  >>  That's very kludgy, isn't it ?

  brogoff> IMO, it's inconvenient, and a drawback to the language, but
  brogoff> not a kludge, or even a mistake. I reserve that for the
  brogoff> undefined order of evaluation :-).  Oh well, you didn't
  brogoff> expect paradise, did you?

  >> I'm assuming the right way is to use modules, which is what
  >> started this whole exercise for me.

  brogoff> If you must share field names in the same module, your only
  brogoff> alternative is to use the OO part of Ocaml. That's
  brogoff> unfortunate, because I think objects are a bit heavyweight
  brogoff> for things like points and segments in a computational
  brogoff> geometry program. OTOH, Ocaml's object system is quite
  brogoff> powerful and principled (and sometimes complex and
  brogoff> unintuitive!) compared to lesser languages like C++ and
  brogoff> Java. I often think a class system like CLOS or Dylan's
  brogoff> would be more to my liking, but hard to reconcile with
  brogoff> ststic typing and modules.

I do _not_ need to use the same field name in the same module.
So using modules is the right way and what I will do.  

For my current application objects are, as you say, too heavyweight.

Brian

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