[Caml-list] Run time error on exception

[Caml-list] Run time error on exception

[Lorenzo Pomili]

[-- Attachment #1: Type: text/plain, Size: 1059 bytes --]

Friend of mine, pointed out to me a case in Reason who cause no compile
time error but run time error, I had reimplemented similar case in OCaml:



exception Generic_exc;;

let string_of_string s = string_of_int (int_of_string s)

let foo = string_of_string "foo"

let fun_exeption x =
    match x with
    0 -> 0
    | _ -> raise Generic_exc
;;

let bar = string_of_int (fun_exeption 5)



both foo and bar don't fail at compile time fail at run time, me and my
friend also try to reimplement similar case (the first one with string and
int) in Elm, and Elm compiler just say you have to manage exeption, because
function can actualy return exeption instead of int.

There is a compile params or similar for check this kind of error? is a
side effect of have a multi-paradigm? and if is it, how to manage it?

I belive I can use OCaml in functional way without just making attention to
not use side effect but if use of function like int_of_string can make run
time error, I probably approac to OCaml in wrong way, and I wanna
understand how to correct it.

[-- Attachment #2: Type: text/html, Size: 1186 bytes --]

Re: [Caml-list] Run time error on exception

[Yawar Amin]

[-- Attachment #1: Type: text/plain, Size: 2673 bytes --]

Hi Lorenzo,

On Wed, Jan 31, 2018 at 4:14 PM, Lorenzo Pomili <lorenzo.pomili85@gmail.com>
wrote:

> Friend of mine, pointed out to me a case in Reason who cause no compile
> time error but run time error, I had reimplemented similar case in OCaml:
>

Runtime errors are very much possible in OCaml/Reason, OCaml does not make
any guarantees about runtime errors in general. The only guarantee OCaml
makes is that you won't get runtime *type* errors (with a couple of known
exceptions). In other words, if your program compiles (passes typechecking)
it won't throw a type error when it runs.

exception Generic_exc;;
>
> let string_of_string s = string_of_int (int_of_string s)
>
> let foo = string_of_string "foo"
>
> let fun_exeption x =
>     match x with
>     0 -> 0
>     | _ -> raise Generic_exc
> ;;
>
> let bar = string_of_int (fun_exeption 5)
>
> both foo and bar don't fail at compile time fail at run time,
>

Yeah, it makes sense that they don't fail at compile time, because they
pass typechecking according to the OCaml typing rules.


> me and my friend also try to reimplement similar case (the first one with
> string and int) in Elm, and Elm compiler just say you have to manage
> exeption, because function can actualy return exeption instead of int.
>

Elm tries to disallow *all* runtime errors (not just runtime type errors),
so it uses various techniques like using the Maybe and Result types to
indicate failed operations. And it makes you handle the possibility of
failure up-front. This is really nice, and you can also do this in OCaml,
but generally it is opt-in, e.g. you can use List.find (which raises an
exception if it doesn't find the item) or List.find_opt (which returns an
option containing the result of the list search).

There is a compile params or similar for check this kind of error? is a
> side effect of have a multi-paradigm? and if is it, how to manage it?
>

You have to manage it manually, and use various techniques to help. For
example, there's a nice pattern-matching technique you can use to return
values even after handling exceptions:

    let bar = string_of_int (match fun_exeption 5 with
    | int -> int
    | exception Generic_exc -> 0)


> I belive I can use OCaml in functional way without just making attention
> to not use side effect but if use of function like int_of_string can make
> run time error, I probably approac to OCaml in wrong way, and I wanna
> understand how to correct it.
>

OCaml is a mix between safety and pragmatism, it doesn't take the extreme
safety approach that Elm does. There are ups and downs to both approaches,
you would have to decide which is best for you.

Regards,

Yawar

[-- Attachment #2: Type: text/html, Size: 3968 bytes --]

Re: [Caml-list] Run time error on exception

[Malcolm Matalka]

Exceptions basically mean "runtime error".  This is why some APIs use
the result type instead of exceptions.  So the API could be something
like:

val int_of_string : string -> (int, err) result

where err is some error type.

I wrote a blog post about this years ago, the APIs in it might be out
dated but the general concept is probably still applies:

http://functional-orbitz.blogspot.se/2013/01/experiences-using-resultt-vs-exceptions.html

Lorenzo Pomili <lorenzo.pomili85@gmail.com> writes:

> Friend of mine, pointed out to me a case in Reason who cause no compile
> time error but run time error, I had reimplemented similar case in OCaml:
>
>
>
> exception Generic_exc;;
>
> let string_of_string s = string_of_int (int_of_string s)
>
> let foo = string_of_string "foo"
>
> let fun_exeption x =
>     match x with
>     0 -> 0
>     | _ -> raise Generic_exc
> ;;
>
> let bar = string_of_int (fun_exeption 5)
>
>
>
> both foo and bar don't fail at compile time fail at run time, me and my
> friend also try to reimplement similar case (the first one with string and
> int) in Elm, and Elm compiler just say you have to manage exeption, because
> function can actualy return exeption instead of int.
>
> There is a compile params or similar for check this kind of error? is a
> side effect of have a multi-paradigm? and if is it, how to manage it?
>
> I belive I can use OCaml in functional way without just making attention to
> not use side effect but if use of function like int_of_string can make run
> time error, I probably approac to OCaml in wrong way, and I wanna
> understand how to correct it.

Re: [Caml-list] Run time error on exception

[Lorenzo Pomili]

[-- Attachment #1: Type: text/plain, Size: 1954 bytes --]

Thanks for reply, one more question there is a way in OCaml, for know if
function can produce side effects? I mean automatic way not read
documentation or source code ^^"

2018-02-01 8:10 GMT+01:00 Malcolm Matalka <mmatalka@gmail.com>:

> Exceptions basically mean "runtime error".  This is why some APIs use
> the result type instead of exceptions.  So the API could be something
> like:
>
> val int_of_string : string -> (int, err) result
>
> where err is some error type.
>
> I wrote a blog post about this years ago, the APIs in it might be out
> dated but the general concept is probably still applies:
>
> http://functional-orbitz.blogspot.se/2013/01/experiences-using-resultt-vs-
> exceptions.html
>
> Lorenzo Pomili <lorenzo.pomili85@gmail.com> writes:
>
> > Friend of mine, pointed out to me a case in Reason who cause no compile
> > time error but run time error, I had reimplemented similar case in OCaml:
> >
> >
> >
> > exception Generic_exc;;
> >
> > let string_of_string s = string_of_int (int_of_string s)
> >
> > let foo = string_of_string "foo"
> >
> > let fun_exeption x =
> >     match x with
> >     0 -> 0
> >     | _ -> raise Generic_exc
> > ;;
> >
> > let bar = string_of_int (fun_exeption 5)
> >
> >
> >
> > both foo and bar don't fail at compile time fail at run time, me and my
> > friend also try to reimplement similar case (the first one with string
> and
> > int) in Elm, and Elm compiler just say you have to manage exeption,
> because
> > function can actualy return exeption instead of int.
> >
> > There is a compile params or similar for check this kind of error? is a
> > side effect of have a multi-paradigm? and if is it, how to manage it?
> >
> > I belive I can use OCaml in functional way without just making attention
> to
> > not use side effect but if use of function like int_of_string can make
> run
> > time error, I probably approac to OCaml in wrong way, and I wanna
> > understand how to correct it.
>
>

[-- Attachment #2: Type: text/html, Size: 2794 bytes --]

Re: [Caml-list] Run time error on exception

[Malcolm Matalka]

Lorenzo Pomili <lorenzo.pomili85@gmail.com> writes:

> Thanks for reply, one more question there is a way in OCaml, for know if
> function can produce side effects? I mean automatic way not read
> documentation or source code ^^"

Nope.  At least not without some third-party tooling, I'm not aware of
any.

>
> 2018-02-01 8:10 GMT+01:00 Malcolm Matalka <mmatalka@gmail.com>:
>
>> Exceptions basically mean "runtime error".  This is why some APIs use
>> the result type instead of exceptions.  So the API could be something
>> like:
>>
>> val int_of_string : string -> (int, err) result
>>
>> where err is some error type.
>>
>> I wrote a blog post about this years ago, the APIs in it might be out
>> dated but the general concept is probably still applies:
>>
>> http://functional-orbitz.blogspot.se/2013/01/experiences-using-resultt-vs-
>> exceptions.html
>>
>> Lorenzo Pomili <lorenzo.pomili85@gmail.com> writes:
>>
>> > Friend of mine, pointed out to me a case in Reason who cause no compile
>> > time error but run time error, I had reimplemented similar case in OCaml:
>> >
>> >
>> >
>> > exception Generic_exc;;
>> >
>> > let string_of_string s = string_of_int (int_of_string s)
>> >
>> > let foo = string_of_string "foo"
>> >
>> > let fun_exeption x =
>> >     match x with
>> >     0 -> 0
>> >     | _ -> raise Generic_exc
>> > ;;
>> >
>> > let bar = string_of_int (fun_exeption 5)
>> >
>> >
>> >
>> > both foo and bar don't fail at compile time fail at run time, me and my
>> > friend also try to reimplement similar case (the first one with string
>> and
>> > int) in Elm, and Elm compiler just say you have to manage exeption,
>> because
>> > function can actualy return exeption instead of int.
>> >
>> > There is a compile params or similar for check this kind of error? is a
>> > side effect of have a multi-paradigm? and if is it, how to manage it?
>> >
>> > I belive I can use OCaml in functional way without just making attention
>> to
>> > not use side effect but if use of function like int_of_string can make
>> run
>> > time error, I probably approac to OCaml in wrong way, and I wanna
>> > understand how to correct it.
>>
>>

Re: [Caml-list] Run time error on exception

[Yawar Amin]

[-- Attachment #1: Type: text/plain, Size: 555 bytes --]

Hi Lorenzo,

On Sun, Feb 4, 2018 at 1:14 PM, Lorenzo Pomili <lorenzo.pomili85@gmail.com>
wrote:

> Thanks for reply, one more question there is a way in OCaml, for know if
> function can produce side effects? I mean automatic way not read
> documentation or source code ^^"
>

Currently no. But there is ongoing research and development on introducing
algebraic effect handlers to OCaml, which will mark functions at the type
level (if I understand correctly) as effectful:
http://kcsrk.info/ocaml/multicore/2015/05/20/effects-multicore/

Regards,

Yawar

[-- Attachment #2: Type: text/html, Size: 1043 bytes --]