(j3.2006) (SC22WG5.5768) RE: Counter argument to the "slow down" sentiment

[Whitlock, Stan]

What counter argument?  You have one unqualified quote.  We can only assume that "all this" refers to enhancing the Fortran Standard.

The Fortran customers I have associated with over the last 40 years of compiler development have overwhelmingly balked at jumping into new changes.  We're shipping Intel Fortran 17.0 but have customers still using 15.0 or earlier - 2 or more years old.  Why?  Because it's extremely expensive and scary to validate a new version of the compiler.  They don't want to do it every year just to get new features.  And why do vendors still support deleted and obsolescent features?  Because customers don't even know if their programs use these old features but they don't want to take the risk so...  "maintain everything" is their demand.

What about specifying new features in the standard and then waiting for the vendors to get around to implementing them?  I mention PDTs and UDIO!  Interps on these were filed mostly by vendors as they were trying to figure out how to implement the features.  Up until that point, no interps because no one found the bugs.  The longer a spec sits unimplemented, the more it rots.

Any guesses on how long it will take the major vendors of Fortran {there used to be dozens...  now how many?} to implement completely the new coarray TS?  Why do you think that F2015 will get implemented enough in the industry to make portability a reality any quicker than F90 or F2003 or F2008?

Too many new features in the standard slow down development - that's what we have seen.  It will continue to be true.

/Stan

PS:  "Dare Mighty Things" is different from "Dare Too Many Things".  WG5's motto needs to be "Dare a Few Mighty Things".

-----Original Message-----
From: j3-bounces at mailman.j3-fortran.org [mailto:j3-bounces at mailman.j3-fortran.org] On Behalf Of Van Snyder
Sent: Saturday, July 09, 2016 6:05 PM
To: sc22wg5 <sc22wg5 at open-std.org>
Subject: (j3.2006) (SC22WG5.5766) Counter argument to the "slow down" sentiment

In a message on comp-fortran-90, a correspondent made this remark:

        Unless and until the pace of change can be accelerated and newer
        and better features can be introduced quickly and made
        available, keep in mind this all appears rather moot for
        industry.

I understand the part about "and made available" depends upon available resources, and each vendor's priorities.  But if we delay specifying new features, they don't get implemented at all, rather than being implemented slowly and at different times by each vendor.  I prefer to have the specs laid down in the standard, not waiting in the wings until every vendor agrees the feature would be near the top of their individual customer-base priority list.

I work for an organization whose motto is "Dare Mighty Things."  If we don't dare anything, we don't accomplish anything.