(j3.2006) (SC22WG5.5734) Minutes
Thu Jun 23 12:04:16 EDT 2016
Here is a revised set of minutes and a diff, following comments from me,
Bill, Makki, and Van. Thanks, David for doing this.
Please let me know by 27 June of any further changes you would like.
-------------- next part --------------
< ISO/IEC JTC1/SC22/WG5 N2109-2
> ISO/IEC JTC1/SC22/WG5 N2109-3
< dinner at the Montana Grill restaurant on the Wednesday evening.
> dinner at Ted's Montana Grill restaurant on the Wednesday evening.
< revised standard at the SC22 meeting in September. The revised JTC1
< Directives would require a letter ballot. The new directives would
< also require wording changes in the document.
> revised standard at the SC22 meeting in September.
> Revised ISO Directives would require wording and format changes in the
< The Further Coarray TS having been published, there were currently no
< contributing development bodies.
> TS 18508 "Additional Parallel Features in Fortran" having been
> published, there were currently no contributing development bodies.
< (N2106) and the WG5 Report to SC22 for 2016 (N2017) were available and
> (N2106) and the WG5 Report to SC22 for 2016 (N2107) were available and
< event at the Montana Grill, Boulder.
> event at Ted's Montana Grill, Boulder.
< Members were invited to examine a draft of the resolutions, N2108-2.
> Members were invited to examine a draft of the resolutions.
< review a new draft of the resolutions, N2108-5.
> review a new draft of the resolutions.
-------------- next part --------------
ISO/IEC JTC1/SC22/WG5 N2109-3
Minutes of Meeting of ISO/IEC JTC1/SC22/WG5
Hosted by INCITS/PL22.3, the US Member Body for JTC1/SC22
in Boulder, CO, USA
June 6-10, 2016
List of Participants:
John Reid (JKR Associates, UK) convenor
Dan Nagle (NCAR, USA) PL22.3 chair
Reinhold Bader (Leibniz Supercomputing Centre, Germany and DIN)
Daniel Chen (IBM, Canada)
Tom Clune (NASA, USA)
Malcolm Cohen (NAG, UK)
Bob Corbett (Oracle, USA)
Steve Lionel (Intel, USA)
Bill Long (Cray, USA)
Nick Maclaren (University of Cambridge, UK)
Lorri Menard (Intel, USA)
Toon Moene (Gnu Fortran, Netherlands)
David Muxworthy (British Standards Institution, UK)
Craig Rasmussen (University of Oregon) (Monday to Thursday)
Damian Rouson (Sourcery Inc, USA) (Monday to Thursday)
Anton Shterenlikht (University of Bristol, UK) (Monday to Thursday)
Van Snyder (Caltech/JPL, USA)
Masayuki Takata (Edogawa University, Japan)
Stan Whitlock (Intel, USA)
Danny Abukalam (Codethink, UK) (Wednesday and Thursday)
Keith Bierman (Emeritus member of PL22.3, USA) (Wednesday)
Walt Brainerd (Emeritus member of PL22.3, USA) (Wednesday)
James van Buskirk (Independent Consultant, USA) (Wednesday)
Srinath Vadlamani (Para Tools, Inc) (Friday)
1. This was a joint meeting of SC22/WG5 and INCITS/PL22.3. These
minutes record only the parts of the meeting in WG5 mode.
2. WG5 papers are referenced as Nnnnn. They are available from
3. INCITS/PL22.3 is abbreviated throughout to PL22.3 and its papers
are referenced as J3/16-nnn. They are available from
1. Opening of the Meeting
The meeting opened at 08:30 on Monday, June 6th 2016.
2. Opening business
2.1 Introductory remarks from the Convenor
The convenor welcomed participants, especially Anton Shterenlikht who
was attending WG5 for the first time.
2.2 Welcome from the Host
Dan Nagle welcomed participants on behalf of INCITS and NCAR.
2.3 Local arrangements
Dan Nagle described local arrangements and invited participants to
dinner at Ted's Montana Grill restaurant on the Wednesday evening.
2.4 Appointments for this meeting
The drafting committee was Reinhold Bader, Daniel Chen, Toon Moene,
David Muxworthy (chair), Steve Lionel and Masayuki Takata. David
Muxworthy acted as secretary and John Reid as librarian.
2.5 Adoption of the agenda [N2105]
The agenda was adopted, subject to the amendment that there would be a
30 minute presentation on coarrays on the Tuesday morning.
3. Matters arising from the minutes of the London 2015 Meeting [N2068]
There were no items not otherwise on the agenda.
4. Status of London 2015 Resolutions [N2077]
The action in resolution L10, that the convenor should provide a
document describing the reasons why WG5 did not apply for a new
project to develop a TS on Units of Measure in Fortran, was still to
With regard to resolution L7, the fourth corrigendum for Fortran 2008
had been submitted to ITTF but not yet published. JTC1 procedures
stated that no more than two corrigenda should be published and that
corrigenda should not normally be used for standards more than three
years old. The preferred JTC1 procedure was to issue an updated
standard containing only corrections.
With regard to resolution L11, participants were asked to consider
possible locations for a meeting in 2019.
5.1 SC22 Matters (Convener)
John Reid said he intended to submit a new work item proposal for the
revised standard at the SC22 meeting in September.
Revised ISO Directives would require wording and format changes in the
5.2 National Activity Reports
Canada: SCC had voted to confirm the Fortran base language. They had
noticed that WG5 had not created a work item for the next
standard revision and they had informed WG23 that there
would be a revision of the vulnerability document for
Germany: Reinhold Bader was currently the only DIN member active in
the Fortran area. A user questionnaire at LRZ on current
and future HPC needs showed that on their Tier-1 system
Fortran was still a majority language with respect to
both code bases and cycles; however many codes were using
mixed language programming.
Japan: The Fortran working group in Japan had ten members. It was
still preparing the Japanese version of Fortran 2008. More
time was needed.
Netherlands: The Programming Languages Committee of NEN, which had
members for C, C++, Cobol and Fortran, had been
disbanded. Toon Moene continued as an individual expert
member of WG5.
UK: A report was in N2104.
US: The main activity was that of PL22.3.
5.3 Report from Primary Development Body (INCITS/J3 Chair)
This was covered elsewhere in the agenda.
5.4 Reports from other Development Bodies (Editors/Heads)
TS 18508 "Additional Parallel Features in Fortran" having been
published, there were currently no contributing development bodies.
5.5 Liaison Reports:
NCITS/J11 (C): Dan Nagle
The Cplex effort was reaching maturity and was being forwarded to
the C committees for inclusion in a future C standard. It
appeared to be little more than OpenMP-style parallelism.
MPI: Bill Long
There was to be an MPI Forum meeting in this same week. It would
be mainly in new feature mode for the next revision. Potentially
interesting features were fault-tolerance progress, and InfoHints
standardization for help with optimization.
UPC: Damian Rouson
The UPC language was stable and there was not a great deal of
IFIP/WG2.5: Van Snyder
There was to be a meeting in Oxford in August, to be followed by
OpenMP: Bill Long
The OpenMP language committee were to release a Technical Report
at Supercomputing 2016, and they were preparing changes to the
specification for that TR, and also for the OpenMP 5.0 release to
follow (in 2017 or 2018).
New facilities since February specific to Fortran were:
- allow array sections in reduction clause for Fortran.
- use "allocated" or "unallocated" to refer to allocation status
- clarify usage of array sections with vector subscripts in map,
to, from, depend, and reduction clauses.
- indicate the references to whole arrays in target region is
unspecified if only a partial array section was mapped.
Other changes in the works:
- [CLARIFICATION] If a Fortran pointer appears in a firstprivate
clause, both the dope vector and the target should be privatized.
- [CLARIFICATION] disallow common blocks on depend and linear
- [NEW] allow function pointers in declare target.
Some other topics being discussed were:
- Clarify handling of Fortran pointers/allocatables in target,
target data, and target update constructs.
- Whether to clarify when compiler may generate copyin/copyout
code at callsites where shared variables are passed in. The
committee thought keeping this implementation-defined was
appropriate (they believe this to be an issue to be clarified
by Fortran, not OpenMP), and that users should consult their
vendor as necessary.
- Interaction between OpenMP memory model and C11/C++11 memory
model. Discussion on interaction with Fortran coarrays memory
model would be deferred (the committee was focusing on
completing F2003 support for 5.0).
WG9 (Ada): Van Snyder
There was to be a meeting the following week. There was a
proposal to revise Fortran interoperability.
WG23 (Vulnerabilities): Dan Nagle
The draft Fortran Annex needed to be updated to accommodate
newly defined vulnerabilities. This work was proceeding but
with low priority.
6. Review of the current draft of the next revision.
More information about the J3