From owner-sc22wg5+sc22wg5-dom9=www.open-std.org@open-std.org Thu Sep 19 04:04:37 2024 Return-Path: X-Original-To: sc22wg5-dom9 Delivered-To: sc22wg5-dom9@www.open-std.org Received: by www.open-std.org (Postfix, from userid 521) id 5C94F357045; Thu, 19 Sep 2024 04:04:37 +0200 (CEST) Delivered-To: sc22wg5@open-std.org Received: from nag-j.co.jp (bvdeuz19.secure.ne.jp [180.222.80.19]) by www.open-std.org (Postfix) with SMTP id A2713356FA7 for ; Thu, 19 Sep 2024 04:04:34 +0200 (CEST) Received: (qmail 78123 invoked from network); 19 Sep 2024 11:04:31 +0900 Received: from unknown (HELO Maru10) (218.42.159.105) by 0 with SMTP; 19 Sep 2024 11:04:31 +0900 From: "Malcolm Cohen" To: "'WG5'" References: <20240918092035.40F30357023@www.open-std.org> In-Reply-To: Subject: RE: [J3] [SC22WG5.6608] RE: RE: [ukfortran] intent(in) and value Date: Thu, 19 Sep 2024 11:04:30 +0900 Message-ID: <002301db0a38$44312780$cc937680$@nag-j.co.jp> MIME-Version: 1.0 Content-Type: multipart/alternative; boundary="----=_NextPart_000_0024_01DB0A83.B41D8A70" X-Mailer: Microsoft Outlook 16.0 Thread-Index: AQH/9fLaKj7MZ/xChKmAc+pwsbYi4AL6IgZnAXKjyD4C2KYDnrHaWKKg Content-Language: ja Sender: owner-sc22wg5@open-std.org Precedence: bulk This is a multipart message in MIME format. ------=_NextPart_000_0024_01DB0A83.B41D8A70 Content-Type: text/plain; charset="us-ascii" Content-Transfer-Encoding: 7bit Yes. INTENT(IN) makes the dummy argument non-definable. This may be useful e.g. so that later on, in program maintenance, someone doesn't try to change the dummy argument in the expectation that it would change the actual argument. Cheers, -- ..............Malcolm Cohen, NAG Oxford/Tokyo. From: J3 On Behalf Of Klemm, Michael via J3 Sent: Thursday, September 19, 2024 10:34 AM To: Jeff Hammond ; General J3 interest list ; 'WG5' Cc: Klemm, Michael ; Malcolm Cohen Subject: [J3] [SC22WG5.6608] RE: RE: [ukfortran] intent(in) and value Hi folks, Thanks for the clarification! One more thing: Is it then sensible to have a variable have the attributes VALUE and INTENT(IN) at the same time? From: Jeff Hammond > Sent: Wednesday, September 18, 2024 7:29 PM To: General J3 interest list >; 'WG5' > Cc: Malcolm Cohen > Subject: Re: [J3] [SC22WG5.6606] RE: [ukfortran] intent(in) and value Caution: This message originated from an External Source. Use proper caution when opening attachments, clicking links, or responding. Thanks. This is useful context and a helpful explanation. Jeff From: J3 > on behalf of Malcolm Cohen via J3 > Date: Wednesday, 18. September 2024 at 12.20 To: 'WG5' > Cc: Malcolm Cohen > Subject: [J3] [SC22WG5.6606] RE: [ukfortran] intent(in) and value External email: Use caution opening links or attachments Hi Jeff, It is called "unspecified INTENT" not "default INTENT". That is because it is not a default, it is that it is not specified. (ASIDE: I don't think we use the phrase in the standard itself, but we do when talking about it.) R828 by itself does not define the INTENT attribute, it defines some syntax - and that is the syntax for specified intent (it's even called "intent-spec". The whole of 8.5.10 defines the INTENT attribute, and indeed, there is a paragraph about unspecified intent. INTENT(INOUT) is very different indeed from unspecified intent. In particular, INOUT requires that the actual argument be a definable variable. There is no such requirement for unspecified intent. With unspecified intent you simply have to get it right, viz not reference the dummy argument unless it is defined, and not define the dummy argument unless it is definable. This is the world before the invention of INTENT. To summarise: with no intent-spec, the intent is not "INOUT", or "INOUT except stuff", it is "unspecified", and has the same capabilities and limitations that it had before Fortran 90. Cheers, -- ..............Malcolm Cohen, NAG Oxford/Tokyo. From: owner-sc22wg5@open-std.org > On Behalf Of Jeff Hammond Sent: Wednesday, September 18, 2024 4:12 PM To: WG5 > Subject: [ukfortran] [SC22WG5.6604] intent(in) and value 8.5.10 R828 says that INTENT is IN, OUT or INOUT. 8.5.19 C868 says that VALUE cannot be INTENT(OUT) or INTENT(INOUT) Should I understand that dummy arguments without an INTENT do not have an implicit intent corresponding to any of the explicit INTENT types? I have always assumed that the default INTENT was INOUT, but that cannot be true since that would conflict with VALUE in the example below. It seems the default INTENT is "INOUT, unless VALUE, in which case it's IN-but-also-assignable". subroutine foo(a,b) implicit none integer, value, intent(in) :: a integer, value :: b ! integer, value, intent(inout) :: c ! not allowed ! integer, value, intent(out) :: d ! not allowed ! a = 21 ! can't assign because intent(in) b = 23 ! can assign end subroutine foo Thanks Jeff ------=_NextPart_000_0024_01DB0A83.B41D8A70 Content-Type: text/html; charset="us-ascii" Content-Transfer-Encoding: quoted-printable

Yes.

 

INTENT(IN) = makes the dummy argument non-definable. This may be useful e.g. so that = later on, in program maintenance, someone doesn’t try to change = the dummy argument in the expectation that it would change the actual = argument.

 

Cheers,<= /o:p>

-- =

.............= .Malcolm Cohen, NAG Oxford/Tokyo.

 

From:<= /b> J3 = <j3-bounces@mailman.j3-fortran.org> On Behalf Of Klemm, = Michael via J3
Sent: Thursday, September 19, 2024 10:34 = AM
To: Jeff Hammond <jehammond@nvidia.com>; General J3 = interest list <j3@mailman.j3-fortran.org>; 'WG5' = <sc22wg5@open-std.org>
Cc: Klemm, Michael = <Michael.Klemm@amd.com>; Malcolm Cohen = <malcolm@nag-j.co.jp>
Subject: [J3] [SC22WG5.6608] RE: = RE: [ukfortran] intent(in) and value

 

Hi = folks,

 

Thanks for the = clarification!

 

One more = thing: Is it then sensible to have a variable have the attributes VALUE = and INTENT(IN) at the same time?

 

From:<= /b> Jeff = Hammond <jehammond@nvidia.com> =
Sent: Wednesday, September 18, 2024 7:29 PM
To: = General J3 interest list <j3@mailman.j3-fortran.org&g= t;; 'WG5' <sc22wg5@open-std.org>
C= c: Malcolm Cohen <malcolm@nag-j.co.jp>
Sub= ject: Re: [J3] [SC22WG5.6606] RE: [ukfortran] intent(in) and = value

 

Caution: This message originated from = an External Source. Use proper caution when opening attachments, = clicking links, or responding. =

 

Thanks.  This is useful context and a = helpful explanation.

 

Jeff

 

From: J3 <j3-bounces@mailman.j3-f= ortran.org> on behalf of Malcolm Cohen via J3 <j3@mailman.j3-fortran.org&g= t;
Date: Wednesday, 18. September 2024 at 12.20
To: = 'WG5' <sc22wg5@open-std.org>
C= c: Malcolm Cohen <malcolm@nag-j.co.jp>
Sub= ject: [J3] [SC22WG5.6606] RE: [ukfortran] intent(in) and = value

Ex= ternal email: Use caution opening links or attachments =

 

Hi = Jeff,

 =

It is called = “unspecified INTENT” not “default INTENT”. That = is because it is not a default, it is that it is not specified. (ASIDE: = I don’t think we use the phrase in the standard itself, but we do = when talking about it.)

 =

R828 by = itself does not define the INTENT attribute, it defines some syntax = – and that is the syntax for specified intent (it’s even = called “intent-spec”. The whole of 8.5.10 defines the INTENT = attribute, and indeed, there is a paragraph about unspecified = intent.

 =

INTENT(INOUT)= is very different indeed from unspecified intent. In particular, INOUT = requires that the actual argument be a definable variable. There is no = such requirement for unspecified intent. With unspecified intent you = simply have to get it right, viz not reference the dummy argument unless = it is defined, and not define the dummy argument unless it is definable. = This is the world before the invention of INTENT.

 =

To = summarise: with no intent-spec, the intent is not “INOUT”, = or “INOUT except stuff”, it is “unspecified”, = and has the same capabilities and limitations that it had before Fortran = 90.

 =

Cheers,

-- =

.............= .Malcolm Cohen, NAG Oxford/Tokyo.

 =

From:<= /b> owner-sc22wg5@open-std.org= <owner-sc22wg5@open-std.org= > On Behalf Of Jeff Hammond
Sent: Wednesday, = September 18, 2024 4:12 PM
To: WG5 <sc22wg5@open-std.org>
S= ubject: [ukfortran] [SC22WG5.6604] intent(in) and value

 

8.5.10 = R828 says that INTENT is IN, OUT or INOUT.

 

8.5.19 = C868 says that VALUE cannot be INTENT(OUT) or INTENT(INOUT)

 

Should I = understand that dummy arguments without an INTENT do not have an = implicit intent corresponding to any of the explicit INTENT types?  = I have always assumed that the default INTENT was INOUT, but that cannot = be true since that would conflict with VALUE in the example below.  = It seems the default INTENT is “INOUT, unless VALUE, in which case = it’s IN-but-also-assignable”.

 

subroutine foo(a,b)

    implicit none

    integer, value, intent(in) = :: a

    integer, value :: = b

    ! integer, value, = intent(inout) :: c ! not allowed

    ! integer, value, = intent(out) :: d ! not allowed

    ! a =3D 21 ! can't assign = because intent(in)

    b =3D 23 ! can = assign

end = subroutine foo

 

Thanks

 

Jeff

=
------=_NextPart_000_0024_01DB0A83.B41D8A70--