[J3] [SC22WG5.6413] Fortran 202Y Suggestions

[Steve Lionel]

Van,

This is a list of suggestions received, not a list of what the committee
accepts. For most of these there is a link to the suggestion source - I
suggest you go there to read the rationale. For the particular suggestion
you mention, the intent is to remove implicit typing based on the first
letter of the entity name. There are papers on the J3 site (181r1, 182)
from two subgroups with initial comments.

We will be doing further discussion of the suggestions in the J3 member
discussion board/forum (the space for this is not yet set up.) As you are a
J3 member (Emeritus) you are welcome to participate there.

We're aware that a cpp-like preprocessor was proposed before. A
preprocessor in no way replaces generics/templates, nor vice-versa. There
are papers from the current meeting detailing the desired semantics of
templates, which have been in design for two years now - note that any
syntax in the papers is for illustrative purposes only.

Steve

On Fri, Jul 22, 2022 at 3:47 PM Van Snyder via J3 <j3 at mailman.j3-fortran.org>
wrote:

> On Fri, 2022-07-22 at 06:04 -0600, Steve Lionel via J3 wrote:
>
> I have updated J3 document 22-176 to r5 (
> https://j3-fortran.org/doc/year/22/22-176r5.pdf) to reflect additional
> submissions of suggested features/changes and to note the three features
> deferred by WG5 to this revision.
>
> Subgroups are sorting the list, deciding which to recommend for inclusion.
> There's been good discussion on many of these already and we have a great
> start on the next revision. Subgroups may choose to accept additional
> submissions over the next year. My hope is to have a near-final list by the
> end of the 2023 joint meeting.
>
> Thanks to all who contributed.
>
> Steve
>
>
> I don't understand the desired effect of "delete default implicit typing."
> Is it the intent that if
>
>   integer :: x = 5
>
> appears in a procedure, and there is no explicit
>
>   save :: x
>
> then the declaration becomes executable?
>
> I have been in favor of default initial values, including for absent
> optional arguments, but this method, if that's what it is, is confusing. I
> would prefer something more like
>
>   integer, automatic :: x = 5
>
> which would not break existing codes that depend upon initialization
> confering the SAVE attribute.
>
> There is also a proposed work item to standardize a cpp-like preprocessor.
> That was introduced to the work plan in 1997 for the 2000 Standard, and
> carried forward as Part III. It was so successful and so widely used that
> it was withdrawn.
>
> Is a cpp-like preprocessor really needed if we do generics right?
>
> If we do generics, and then add macros (which are complementary, not
> exclusive -- there are things that each can do that the other can't), would
> a cpp-like preprocessor provide any additional utility?
>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://mailman.j3-fortran.org/pipermail/j3/attachments/20220722/b8dd840d/attachment.htm>