(j3.2006) (SC22WG5.5535) WG5 straw ballot on draft TS
John Reid
John.Reid
Fri Aug 14 13:09:02 EDT 2015
Dear all,
Bill Long has made all the changes to the TS that we agreed in London
and made a small number of additional editorial changes. The new version
is attached (N2074) with Bill's explanation of everything he has done
(N2075). I am now starting a WG5 straw ballot (N2079) on whether this
draft is suitable for forwarding for publication. The deadline is 12 Sept.
Thanks, Bill, for doing this so quickly.
John.
-------------- next part --------------
ISO/IEC JTC1/SC22/WG5 N2079
WG5 straw ballot on N2074
John Reid, 14 August 2015
This is a WG5 straw ballot on N2079, the draft TS 18508,
Additional Parallel Features in Fortran, following the successful
ballot on the PDTS, see N2065.
N2079 was prepared by the editor, Bill Long, and incorporates all the
changes to the PDTS made in response to the comments in the ballot, see
N2066. Details of the changes made from the previous draft, N2056, are
given in N2075.
In the final version sent to ISO, the page numbers will be turned off.
ISO does not like these, but Bill left them in this version so that it
would be easier for people to make comments.
Please answer the following question "Is N2074 suitable for forwarding to
ISO for publication?" in one of these ways.
1) Yes.
2) Yes, but I recommend the following changes.
3) No, for the following reasons.
4) Abstain.
This is an individual vote. Please send your vote to sc22wg5 at open-std.org
to arrive by 9 a.m. (UK time) on 12 September 2015.
-------------- next part --------------
A non-text attachment was scrubbed...
Name: N2074.pdf
Type: application/pdf
Size: 333253 bytes
Desc: not available
Url : http://mailman.j3-fortran.org/pipermail/j3/attachments/20150814/94ed3f29/attachment-0001.pdf
-------------- next part --------------
ISO/IEC JTC1/SC22/WG5 N2075
Editor's report for WG5/N2074
Bill Long, 13 August 2015
Document WG5/N2074 was prepared by applying edits to WG5/N2056 based
on these papers passed at PL22.3 meeting 207: J3/15-191, J3/15-181r3,
J3/15-196, J3/15-198, J3/15-201, J3/15-205, and J3/15-212r1. Edits
from these papers are prefixed by the paper number with the "J3/15-"
removed. Additional edits are noted by the prefix "Ed:". Comments
following edits are enclosed in { }. The edits are ordered here based
on page and line number of WG5/N2056.
Edits to WG5/N2056:
===================
Edits for Clause 1 Scope
------------------------
212r1:[1:6-8] In 1 Scope, in para2, make these 3 edits: In the first
sentence replace "or progress models" by "model". Delete the second
sentence "Some level ... clauses 7 and 8." In the last sentence
replace "and progress models" by "model".
Edits for Clause 2 Normative References
---------------------------------------
(none)
Edits for Clause 3 Terms and definitions
----------------------------------------
(none)
Edits for Clause 4 Compatibility
--------------------------------
(none)
Edits for Clause 5 Teams of images
----------------------------------
191:[9:5] In 5.1 Team concepts, para 1, sentence 2, replace "both
sibling and ancestor" by "sibling or ancestor".
191:[9:27] In 5.2 TEAM_TYPE, para 1 sentence 3, replace "default
initialized" by "default-initialized".
196:[10:17-18] In 5.3 CHANGE TEAM, move the two sentences of para 3,
following the constraints, "A coselector name ... statement begins
execution." to be the first two sentences of the para 5 that currently
begins "A <codimension-decl> in a <coarray-association>..."
191:[10:19-20] In 5.3 CHANGE TEAM construct, para 2 after constraint
list, delete the first sentence "The value of <team-variable>
... initial team."
{Redundant with the sentence that follows.}
191:[11:13,17] In 5.4 Image selectors, after the constraint, para 1,
last sentence, and para 2, sentence 2, replace "an image index" by
"the image index".
{Use same article as in the first sentence of subclause 6.6 of ISO/IEC
1539-1:2010.}
Edits for Clause 6 Failed images
--------------------------------
Ed:[15:9] In 6.1 Introduction to Failed Images, para 3, replace
"non-failed" by "nonfailed".
{Remove hyphen after "non".}
191:[16:20] In 6.4 STAT_FAILED_IMAGE, para 1, sentence 2, replace
"image image" by "image".
Edits for Clause 7 Events
-------------------------
198:[17:4+] Add a new paragraph after 7.1p1 (Events/Introduction):
"Segments ordered by EVENT POST and EVENT WAIT are included in the
partial order specified in subclause 2.3.5 of ISO/IEC 1539-1:2010."
191:[17:6-7] In 7.2 EVENT_TYPE, para 1, sentence 3, replace "Each
component is fully default initialized." by "Each nonallocatable
component is fully default-initialized.".
191:[17:8] In 7.2 EVENT_TYPE, para 2, sentence 1, replace "An event
variable has a count that" by "The value of an event variable includes
its event count, which".
198:[17:26+] Add a new Note at the end of 7.2 EVENT_TYPE:
"NOTE:
The updates of atomic variables are coherent but not necessarily
consistent, so a processor might have to use extra synchronization
to obtain the consistency required for the segments ordered by
EVENT POST and EVENT WAIT statements."
191:[18:1] In 7.3 EVENT POST statement, in para 2 after the constraint,
replace "the count does not change" by "the value of the count of the
event variable is processor dependent.
{Ed: Omitted the redundant "." at the end of the replacement text in
191.}
191:[18:2-4] In 7.3 EVENT POST statement, delete the final paragraph.
{Effectively only says that actions that occur in unordered segments
are unordered.}
196:[18:4+] In 7.3 EVENT POST, replace Note 7.2 by "The completion of
an EVENT POST statement does not depend on the execution of a
corresponding EVENT WAIT statement".
{Ed: Added the missing "." at the end of the replacement sentence.}
191:[18:20+] In 7.4 EVENT WAIT statement, following the 3-item
list. insert a new paragraph: "If an error occurs during execution of
an EVENT WAIT statement, the value of the count of its event variable
is processor dependent."
196:[18:21-24] In 7.4 EVENT WAIT, replace the para following the
itemized list, beginning "If an EVENT WAIT..." by "An EVENT POST
statement execution is initially unsatisfied. Successful execution of
an EVENT WAIT statement with a threshold of <k> satisfies the first
<k> unsatisfied EVENT POST statement executions for that event
variable. This EVENT WAIT statement execution causes the segment
following the EVENT WAIT statement execution to succeed the segments
preceding those <k> EVENT POST statement executions."
196:[18:25] In 7.4 EVENT WAIT, delete the final sentence, "A failed
image...statement.", before the first Note.
196:[18:25++] In 7.4 EVENT WAIT, delete the second note, Note 7.4.
Edits for Clause 8 Intrinsic procedures
---------------------------------------
212r1:[19:22+] In 8.2 Atomic subroutines, after para 1, add paragraph:
"Atomic operations shall make asynchronous progress (3.2). If the
atomic variable X[A] is defined by an atomic subroutine on image B,
image C repeatedly references X[A] by an atomic subroutine in an
unordered segment, and no other image defines X[A], image C will
eventually receive the value defined by image B, even if none of the
images A, B, or C execute an image control statement until after X[A]
is defined by image B and that value is received by image C."
201:[20:27++] In 8.3 Collective subroutines, add following new
paragraph at the end of NOTE 8.4:
"If the actual argument is a coarray, the implementation might be able
to optimize the execution of the collective, for example by avoiding
copying of data."
181r3:[26:30-32] In 8.4.13 CO_REDUCE, replace the Example text "If the
number of images ... on both images" by
"The subroutine below demonstrates how to use CO_REDUCE to create
a collective counterpart to the intrinsic function ALL:
SUBROUTINE co_all(boolean)
LOGICAL, INTENT(INOUT) :: boolean
CALL CO_REDUCE(boolean,both)
CONTAINS
PURE FUNCTION both(lhs,rhs) RESULT(lhs_and_rhs)
LOGICAL, INTENT(IN) :: lhs,rhs
LOGICAL :: lhs_and_rhs
lhs_and_rhs = lhs .AND. rhs
END FUNCTION both
END SUBROUTINE co_all
"
205:[26:38-43] In 8.4.14 CO_SUM, in the description of the argument A,
replace the last two sentences "If it is a scalar ... all
corresponding elements in the current team." by
"If it is a scalar, the computed value is equal to a
processor-dependent approximation to the sum of the values of A on all
images in the current team. If it is an array, it shall have the same
shape on all images in the current team and each element of the
computed value is equal to a processor-dependent approximation to the
sum of all corresponding elements of A on the images in the current
team. If RESULT_IMAGE is present, the computed value may depend on
the image that the RESULT_IMAGE specifies; otherwise, the computed
value is the same on all images in the current team."
{Ed: Deleted the "the" before the last "RESULT_IMAGE".}
196:[27:15] In 8.4.15 EVENT_QUERY, in the description of the EVENT
argument, after "ISO_FORTRAN_ENV." insert a new sentence "It shall not
be coindexed."
{Disallowing a coindexed event argument in EVENT_QUERY makes it
consistent with EVENT WAIT.}
191:[27:17] In 8.4.15 EVENT_QUERY, in the description of COUNT, change
"that that" to "than that".
191:[27:18] In 8.4.15 EVENT_QUERY, in the description of COUNT, change
"conditions" to "condition".
191:[27:19] In 8.4.15 EVENT_QUERY, at the end of the description of
COUNT change "assigned the value 0" to "assigned the value -1". {Make
the value returned in the case of an error different from any value
that is possible when there is no error.}
196:[27:22] In 8.4.15 EVENT_QUERY, in the first sentence of the Example,
after "posts or waits", add "in preceding segments , and for which
there are no posts or waits in an unordered segment".
196:[27:25] In 8.4.15 EVENT_QUERY, in the second block of text of the
Example, replace "EVENT[2]" by "EVENT".
196:[27:26] In 8.4.15 EVENT_QUERY, in the second block of text of the
Example, after "specification" add "in preceding segments, and for
which there are no posts or waits in an unordered segment".
196:[27:27] In 8.4.15 EVENT_QUERY, in the second CALL statement of the
Example, replace "EVENT[2]" by "EVENT".
191:[30:39+] In 8.5.2 IMAGE_INDEX, add as para 4 "The result is the
image index in the specified team".
{Ed: Added a final "." that is missing in the edit.}
191:[31:3] In 8.5.3 MOVE_ALLOC, para 1, replace "is modified to" by
"is extended to".
191:[31:4] In 8.5.3 MOVE_ALLOC, para 1, replace "and a modified
result" by "and modified semantics of execution".
191:[31:39+] In 8.5.4 NUM_IMAGES, add as para 4 "The result is the
number of images in the specified team.".
Edits for Clause 9 Required editorial changes to ISO/IEC 1539-1:2010(E)
-----------------------------------------------------------------------
Ed:[42:11] In 9.9 Edits to clause 13, in the second edit for 13.7.165
THIS_IMAGE, in the last sentence of the description of the TEAM
argument, replace "it shall be established for TEAM" by "it shall be
established in the team specified by the value of TEAM".
{Use the same wording as in the following edit, also for THIS_IMAGE.}
Ed:[44:16] In 9.11 Edits to annex A, delete the bullet "how sequences
of event posts in unordered segments interleave with each other
(8.5.2a)".
{The corresponding normative text in the TS was deleted. This concept
is now covered by the processor dependence for atomics, since the
count change caused by an event post is atomic.}
Ed:[44:19-20] In 9.11 Edits to annex A, in the last bullet item,
replace "intrinsic procedure when there is an error condition" by
"intrinsic subroutine when an error condition occurs".
{Make wording parallel to the similar entry in J3/15-007r2 that covers
only atomics. It is intended that this entry replace the one in
J3/15-007r2. The difference is the addition of collective
subroutines.}
191:[44:20+] In 9.11 Edits to annex A, add a new dependency:
"the value of the count of the event variable in an EVENT WAIT or
EVENT POST statement if an error condition occurs (8.5.2a, 8.5.2b);"
{Ed: Subclause numbers added at the end of the edit.}
Ed:[44:20++] In 9.11 Edits to annex A, add a new dependency:
"the value returned by the intrinsic function IMAGE_STATUS if it is
other than zero (13.7.79a);"
{The normative text in 8.4.18 IMAGE_STATUS says the return value can
be processor dependent. But the corresponding entry in this subclause
was omitted.}
Ed:[44:20+++] In 9.11 Edits to annex A, add a new dependency:
"the value assigned to a STAT argument in a reference to the intrinsic
subroutine MOVE_ALLOC when an error condition occurs (13.7.118)."
{The modified MOVE_ALLOC subroutine (8.5.3) now has a STAT argument
and its possible values are processor dependent. This edit was
overlooked when that change was made.}
Edits for Annex A Extended notes
--------------------------------
196:[51:13-14] In A.3.2 EVENT_QUERY example, replace the statements
CALL EVENT_QUERY(submit[i],count,STAT=status) ! Test image i
IF (status==STAT_FAILED_IMAGE) CYCLE
by
IF (IMAGE_STATUS(i) == STAT_FAILED_IMAGE) CYCLE
196:[51:24-25] In A.3.2 EVENT_QUERY example, replace the statements
CALL EVENT_QUERY(submit[i],count,STAT=status) ! Test image i
IF (status==STAT_FAILED_IMAGE) THEN ! Image i has failed
by
IF (IMAGE_STATUS(i) == STAT_FAILED_IMAGE) THEN
196:[51:47-52] In A.3.2 EVENT_QUERY example, replace the lines
CALL EVENT_QUERY(submit[i],count,STAT=status) ! Test image i
IF (status/=STAT_FAILED_IMAGE) &
work_item[i] = create_work_item(kk)
EVENT POST (submit[i],STAT=status)
! If image i has failed, this will not hang and the failure
! will be handled on the next iteration of the loop
by
work_item[i] = create_work_item(kk)
EVENT POST (submit[i],STAT=status)
! If image i has failed, the failure will be handled on
! the next iteration of the master loop.
More information about the J3
mailing list