[J3] work on F202X at #217?

Wed Jul 25 08:40:08 EDT 2018

Apologies - I was using “syntax” in a very vague sense.      I just meant that I wanted to be able to “see” at least in some murky sense how it would work in theory.

I’m a  big believer in process, but a poor disciple.    It took years for someone to bash into my skull what “1. use cases” really means.   Looking forward to trying to better understand the distinctions between 2,3,and 4 on Malcolm’s list.

Cheers,

- Tom

On Jul 25, 2018, at 12:57 AM, Malcolm Cohen via J3 <j3 at mailman.j3-fortran.org<mailto:j3 at mailman.j3-fortran.org>> wrote:

> at least a rough sketch of the proposed syntax

I would prefer to see formal requirements.  In principle the sequence goes:

1. use cases
2. formal requirements
3. formal specifications
4. syntax
5. edits

The last three are usually delegated by WG5 to J3 anyway, but it is the formal requirements that WG5 needs to do.  Although “example proposed syntax” can be useful to focus the mind, WG5 almost never votes on specific syntax anyway; J3 is trusted to “do the right thing” which quite often ends up changing the syntax.

Leaping ahead to syntax, skipping both requirements and specifications, is a bad idea.  Of course playing around with syntax ideas to illuminate the problems, requirements, and specifications, can be very useful, but focussing on syntax too early often leads to later problems.

Of course for trivial features some of these steps are just a single paragraph in a paper that combines lots of steps.  For large proposals intermediate steps may be useful, e.g.
1.5 informal requirements

Formal requirements might sound like pettifogging bureaucracy, but for anything of any size or complexity it is a really good idea.  And yes, they really are different from specifications.

Cheers,
--
..............Malcolm Cohen, NAG Oxford/Tokyo.

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://mailman.j3-fortran.org/pipermail/j3/attachments/20180725/01434793/attachment.html>