0003688: 3 documentation suggestions and 5 camlp4 problems

(0003025)
administrator (administrator)
2005-06-17 14:08

Hi,

About the Camlp4 problems:

> Revised syntax pa_r:
>
> 4. In the revised syntax one can define "type a = ! 'a . list 'a"
> is this really an admissable type?

OCaml forbids such type schemes ($\forall$ ...) at toplevel in its
concrete syntax spec, but allows them as ASTs. The AST produced by
"type a = ! 'a . list 'a" is therefore legal, but should be essentially
unusable.

If it happens to be buggy to allow such top level polymorphic types,
I'll forbid that, then. (Wouldn't be really complex, but wouldn't be
really elegant either.)

> Anyway, pr_o prints it as "type a = 'a . 'a list" which is illegal.

Fixed.

> 5. val is not handled consistenly:
>
> type x = {val : int; kk : int};
> value f a = { (a) with val = 4};
>
> works fine with pa_r but becomes illegal when printed with
> pr_o.

Fixed: val -> contents, and contents -> contents__.
Still buggy if original code already uses "contents__" but well...

> 7. I believe there are severe problems in guessing the
> constructor arity. For instance
>
> type 'a t = A of 'a
> let f = function A(a,b) -> A(b,a)
>
> compiles fine with pa_o but becomes illegal when printed with
> pr_r.

There is no way at parse time of guessing the arity of A: needs type
checking for that.

I've added a -no_curried_constructors option to pr_r: all non constant data
constructors will have arity 1. At least, it should be correct.

> 8. mkumin should match on the other integer types as well.
> Currently <:expr< -5l >>, <:expr< -5L >> and <:expr< -5n >>
> work not as expected in a patterns (they match (~) 5 instead
> of -5).

Fixed.

Sincerely,

-- Michel

(0003026)
administrator (administrator)
2005-10-21 11:34

Fixed the Camlp4 part in June 2005 [MM]

(0007714)
doligez (manager)
2012-07-11 17:15

About documentation:
2: it's in module-expr rather than definition.
3: added it to the "constant" nonterminal

STILL TO FIX: point 1

(0008195)
Kakadu (reporter)
2012-10-03 10:19

Also in OCaml interface file we write

val f : x:int -> y:int -> int

but in revised we should write with ~:

value f : ~x:int -> ~y:int -> int

It seems that Chapter 6 `Revised syntax` doesn't say anything about labeled arguments.

(0008867)
doligez (manager)
2013-02-19 11:42

Point 1: fixed (commit 13301)
All the orignal points are fixed, we are left with the problem raised by Kakadu, so I'm switching this PR to the camlp4 category.

(0008872)
hongboz (developer)
2013-02-19 15:52

It's true that in revised syntax "~" is required for label types, but I did not find a place to add the documentation

Issue History
Date Modified	Username	Field	Change
2005-11-18 10:14	administrator	New Issue
2012-07-11 17:15	doligez	Note Added: 0007714
2012-07-11 17:15	doligez	Category	OCaml general => OCaml documentation
2012-07-11 17:15	doligez	Target Version	=> 4.01.0+dev
2012-07-11 17:15	doligez	Description Updated	View Revisions
2012-07-31 13:37	doligez	Target Version	4.01.0+dev => 4.00.1+dev
2012-09-11 09:53	doligez	Target Version	4.00.1+dev => 4.00.2+dev
2012-09-17 17:57	doligez	Relationship added	has duplicate 0005035
2012-10-03 10:19	Kakadu	Note Added: 0008195
2013-02-19 11:42	doligez	Note Added: 0008867
2013-02-19 11:42	doligez	Category	OCaml documentation => Camlp4
2013-02-19 11:42	doligez	Target Version	4.00.2+dev =>
2013-02-19 15:52	hongboz	Note Added: 0008872

Notes
(0003025) administrator (administrator) 2005-06-17 14:08	Hi, About the Camlp4 problems: > Revised syntax pa_r: > > 4. In the revised syntax one can define "type a = ! 'a . list 'a" > is this really an admissable type? OCaml forbids such type schemes ($\forall$ ...) at toplevel in its concrete syntax spec, but allows them as ASTs. The AST produced by "type a = ! 'a . list 'a" is therefore legal, but should be essentially unusable. If it happens to be buggy to allow such top level polymorphic types, I'll forbid that, then. (Wouldn't be really complex, but wouldn't be really elegant either.) > Anyway, pr_o prints it as "type a = 'a . 'a list" which is illegal. Fixed. > 5. val is not handled consistenly: > > type x = {val : int; kk : int}; > value f a = { (a) with val = 4}; > > works fine with pa_r but becomes illegal when printed with > pr_o. Fixed: val -> contents, and contents -> contents__. Still buggy if original code already uses "contents__" but well... > 7. I believe there are severe problems in guessing the > constructor arity. For instance > > type 'a t = A of 'a > let f = function A(a,b) -> A(b,a) > > compiles fine with pa_o but becomes illegal when printed with > pr_r. There is no way at parse time of guessing the arity of A: needs type checking for that. I've added a -no_curried_constructors option to pr_r: all non constant data constructors will have arity 1. At least, it should be correct. > 8. mkumin should match on the other integer types as well. > Currently <:expr< -5l >>, <:expr< -5L >> and <:expr< -5n >> > work not as expected in a patterns (they match (~) 5 instead > of -5). Fixed. Sincerely, -- Michel

(0003026) administrator (administrator) 2005-10-21 11:34	Fixed the Camlp4 part in June 2005 [MM]

(0007714) doligez (manager) 2012-07-11 17:15	About documentation: 2: it's in module-expr rather than definition. 3: added it to the "constant" nonterminal STILL TO FIX: point 1

(0008195) Kakadu (reporter) 2012-10-03 10:19	Also in OCaml interface file we write val f : x:int -> y:int -> int but in revised we should write with ~: value f : ~x:int -> ~y:int -> int It seems that Chapter 6 `Revised syntax` doesn't say anything about labeled arguments.

(0008867) doligez (manager) 2013-02-19 11:42	Point 1: fixed (commit 13301) All the orignal points are fixed, we are left with the problem raised by Kakadu, so I'm switching this PR to the camlp4 category.

(0008872) hongboz (developer) 2013-02-19 15:52	It's true that in revised syntax "~" is required for label types, but I did not find a place to add the documentation