(j3.2006) (SC22WG5.5396) [ukfortran] Straw vote on draft DTS

[N.M. Maclaren]

On Dec 11 2014, Damian Rouson wrote:
>
> I will probably suffer for asking this, but just out of curiosity, would 
> building such an infrastructure make it more likely that a future 
> standard would support user-specified exception-handling? If so, then 
> from a user perspective, this would be a nice stepping stone.

No.  Less so, if anything.

>> The more general question of whether Fortran should include fault 
>> tolerance on a timely schedule at all is really a question Fortran's 
>> future relevance in the HPC market place.  ....
>
> I find this reasoning very compelling. I've been for some time touting 
> the failed-image feature set as an example of Fortran leading the way in 
> an area that everyone recognizes as important at the exascale.  ...

Have you ever implemented a language run-time system that trapped
arbitrary system exceptions, called back to user code, and allowed
recovery?  Have you ever SEEN one?

Well, I have implemented more than one - and, based on that experience,
this proposal is a non-starter.  It is not practically deliverable.
Back in the mainframe days, a FEW of us delivered a much simpler
variant, but it was very rare indeed for puported recovery mechanisms to
be actually usable.  Since the rise of Unix-derived systems, I have not
seen a single such mechanism - oh, yes, I have seen plenty that claimed
to deliver, but didn't.

The simple fact is that Fortran is leading the way rather like the
fastest lemming - over a cliff.  In 50 years of trying - and I can
assure you that there has been no lack of trying - nobody has EVER
standardised anything like this in an imperative language, or even
specified it at all well for a single system.  That might indicate
something ....

> In my experiences teaching modern Fortran courses, it appears coarrays 
> have breathed new life into a language whose demise had long been 
> rumored.  ...

Mine is quite the opposite.  I am seeing increasing interest in
Fortran, and nobody is showing the slightest interest in coarrays.
They have most of the disadvantages of both MPI and OpenMP.

> Will there be any possibility for implementors to leverage the MPI or 
> SHMEM progress Bill is citing? GFortran/OpenCoarrys is already doing this 
> in other areas: using MPI's one-sided communication and collective 
> communication to support the same in coarray Fortran.

WHAT progress?  I am on the MPI Forum mailing list, and I have seen
none.  I am afraid that Bill is far too keen on citing progress that
has not yet been delivered, or even is purely a glint in the eye of
the people that hope to see it.

>> The idea that vendors need to implement a facility like fault tolerance 
>> before including it in the standard is out of touch with the realities 
>> of modern-day compiler development.  ....

Look, for the Nth time, THERE IS STRONG EVIDENCE THIS CANNOT BOTH BE
DELIVERABLE AND USABLE.  And that is why people like me will never vote
for it until we see some evidence that at least one implementation can
deliver it.


Regards,
Nick Maclaren.