(j3.2006) (SC22WG5.5227) Votes on draft TS
John Reid
John.Reid
Wed Apr 16 05:11:02 EDT 2014
WG5,
Here is the first draft result of our vote. Have I got all the votes and
have I copied them correctly? Please let me know of any errors by 9.a.m.
Friday 18 April.
Best wishes,
John.
-------------- next part --------------
ISO/IEC JTC1/SC22/WG5 N2013-1
Result of the WG5 letter ballot on N2007
John Reid
N2012 asked this question
Please answer the following question "Is N2007 ready for forwarding to
SC22 as the DTS?" in one of these ways.
1) Yes.
2) Yes, but I recommend the following changes.
3) No, for the following reasons.
4) Abstain.
The numbers of answers in each category were:
4 for 1) Yes (Chen, Moene, Nagle, Whitlock)
2 for 2) Yes, but I recommend the following changes (Long, Reid)
4 for 3) No, for the following reasons (Bader, Cohen, Maclaren, Snyder)
2 for 4) Abstain (Corbett, Muxworthy)
Here are the responses in detail
Reinhold Bader
3) No, for the following reasons.
The following items from my vote on the previous draft (see N1999)
have not been dealt with yet:
(5C), items (2),(3),(4),(7)
(6B)
Comments on section A.3.2.2
The conceptual changes in 14-111 (going from team depth to
team variable specification for the purpose of intrinsic
interfaces) have not been discussed in meeting 203.
Furthermore, the following issues have come to my attention:
Section 5:
~~~~~~~~~~
5.1 I would welcome a NOTE at the end of [9:16-24]
that clarifies how dummy arguments are treated:
----------
NOTE 5.1-
A coarray dummy argument is not established in any ancestor
team even if the corresponding actual argument is established
in one or more of them.
----------
5.3 Is it intended that a coarray-association as defined in
[10:24-28] is permitted to use differing cobounds on
each image? If not, should the cobounds be permitted to
differ between teams?
5.4 In [11:11-12] determination of the image index is
presumably based on the local cobounds even though
those may be different from those in the sibling
team (or specifically the targeted image). If the answer
to the questions posed for 5.3 above is "NO", some
additional restriction may be desirable for sibling
team addressing, i.e. the data item must only be
referenced via an associate name.
If the answer to the first question is "YES", adding
a note that describes the above situation is
recommended.
In any case, I consider it useful to extend the
intrinsics IMAGE_INDEX and NUM_IMAGES to enable
appropriate queries on sibling teams.
5.7 I suggest deleting the word "randomly" in the last line of
NOTE 5.6. This is already implied by "chance of failure".
Section 6
~~~~~~~~~
6.4 EVENT WAIT has the problem that it will probably hang forever
if the image that is supposed to do the POST fails. While
this can be ameliorated by
(a) using EVENT_QUERY and FAILED_IMAGES to find out in advance
whether this is the case
(b) making (a) more reliable by dedicating a "monitor" image
to investigate the above (a scenario where EVENT_QUERY on a
remote event would be needed!) and do a bailout POST,
I think adding some semantics to enable a properly defined
exit status is desirable and allows for less klugey fail-safe
programming. For example, one might have an
additional optional argument ERROR_ON=[image-list|*] that
causes the statement to exit with STAT=[STAT_STOPPED_IMAGE|
STAT_FAILED_IMAGE] if something ill happens to an image in
the specified image set. The LOCK, UNLOCK and possibly the
SYNC MEMORY statements suffer from similar problems.
"*" might refer to all images of the current team that have
not failed or been stopped on entry to the statement.
Section 7 Intrinsic procedures
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
7.4.[1-5] All atomic subroutines should permit optional STAT
specification to permit signaling of a failure. I think
it would be OK to specify that this may only work if
the run-time can determine that the coindexed object
that is subject of the atomic operation has become
unavailable and after it has done so; otherwise the
operation is expected to cause image failure itself.
This is necessary to properly support fine-grained
synchronization algorithms that can in principle continue
even when image failures occur. (The example given by me
in N1999/{Comments on section A.3.2.2} would hang if an image
fails). To some extent, this issue is related to 6.4 above
considering that EVENTs might be implemented in terms of
atomics.
7.4.11 The text in [23:31-33] appears outdated given the recent
changes to events. Also the procedure should be made atomic
to make clear it can be invoked concurrently (i.e. in unordered
segments) with updates being made via EVENT POST/WAIT.
Furthermore, it should be made clear that an error can only
arise if the queried event variable is on a failed image.
Annex A:
~~~~~~~~
A.1.4 [42:18] Replace "p**q" by "p*q"
A.2.1 I agree with previous comments that this example uses
EVENT_QUERY incorrectly. An effort of improving the example
also with respect to the fail-safe functionality is under way.
_______________________________________________________________________
Malcolm Cohen
No, the technical content is not yet agreed (see many comments by Nick
Maclaren and Reinhold Bader), and I think it is not correct.
In particular,
(1) the collectives CO MAX, CO MIN, CO REDUCE, CO SUM, should be split
into two forms, one with RESULT, one without. The one with RESULT
should have SOURCE as INTENT(IN), the one without should have SOURCE
as INTENT(INOUT). RESULT must not be optional. The SOURCE INTENT(IN)
form should have no coarray restrictions on SOURCE.
(2) the semantics of EVENT are still unclear. 6.3p3 says
"If the segment that precedes an EVENT POST statement is unordered
with respect to the segment that precedes another EVENT POST statement
for the same event variable, the order of execution of the EVENT POST
statements is processor dependent."
but if they are in unordered segments, there is no order: processor-
dependent or otherwise.
(3) I agree that EVENT WAIT UNTIL_COUNT=-127 makes no sense, but it
seems far more likely to be a progamming error than a deliberate
intention to specify UNTIL_COUNT=1. There should be a requirement
that the <scalar-int-expr> in the UNTIL_COUNT shall be positive so
that processors can be encouraged to report such mistakes.
(4) Note 6.2 should be normative text. Assuming that those are the
semantics we want.
(5) Perhaps Note 6.1 should be normative and state something like
"The segment following execution of an EVENT POST statement is not
ordered with respect to the segment following the corresponding
EVENT WAIT statement.". Mind you that would require us properly
describing the concept of "the corresponding EVENT WAIT" (which we
seem to half-way have, but not clearly described).
(6) Use of EVENT_QUERY + SYNC MEMORY looks very scary. It would be
possible to use this to write remote-event-waiting, I am not entirely
convinced that this is the right thing to do. Perhaps an EVENT POLL
statement would be safer (or an option to EVENT WAIT e.g.
BLOCK=.FALSE.).
(7) And why doesn't EVENT QUERY simply say it returns the event count?
If the event count can ever not be equal to "the number of successful
posts minus the number of successful waits" there is a problem. If it
is by definition equal to "the number of successful posts minus the
number of successful waits", the wording is the problem.
(8) Is the processor supposed to support the following:
INTEGER,PARAMETER :: i4 = SELECTED_INT_KIND(18)
EVENT WAIT(UNTIL_COUNT = HUGE(0_i4))
??? That looks pretty gross so maybe not, but what about
EVENT WAIT(UNTIL_COUNT = 100*1000*1000)
??? (And if that is still gross, EVENT_COUNT=1000 ...)
For that matter, what about many EVENT POSTs before an EVENT WAIT.
Is there a limit? Presumably yes, then let's (a) say there is a
limit (b) require (or at least recommend) the processor to raise an
error condition if the limit is exceeded (c) specify a minimum limit
that the processor must implement; HUGE(0) springs to mind.
_______________________________________________________________________
Bill Long
2) Yes, but I recommend the following change:
At [36:9] change ???5.6??? to ???5.8???.
I agree with at least some of the editorial changes from other ballots,
but will not reproduce them here.
_______________________________________________________________________
Nick Maclaren
3) No, for the following reasons.
Because this is a long response, the most serious and difficult points
are:
Events.2) This TS introduces a facility that is arguably in conflict
with Fortran 2008 1.5 paragraph 1, and is not described as processor
dependent. It is not clear how to fix it, and it should be removed. At
the very least, it should be explicitly specified to be processor-
dependent.
Collectives.1) This TS introduces a conflict with Fortran 2008's concept
and specification of variable definition context. This is definitely
fixable, but may need significant changes to the collectives.
General.1 and General.2) This TS should be explicitly removed from
consideration for inclusion in the next Fortran standard (see N1979),
because it will probably be be infeasible to specify a data consistency
model by the deadline for the first CD ballot, and there is very
unlikely to be sufficient implementation and user experience.
EVENTS
------
Events.1) The wording in 8.7 page 32:26-27 still refers to EVENT POST
and EVENT WAIT being within segments.
8.7 page 32:26-27 should replace:
A coarray that is of type EVENT TYPE may be referenced or defined
during the execution of a segment that is unordered relative to
the execution of another segment in which that coarray of type
EVENT TYPE is defined.
by:
A coarray that is of type EVENT TYPE may be referenced or defined
during the execution of a segment that is unordered relative to the
execution of the EVENT POST and EVENT WAIT statements by which that
coarray is defined.
Events.2) Reinhold Bader's (6B) [lines 258-264] and my Comment I [lines
923-965] in N1999 have still not been addressed. In addition, other
comments along the same lines have been made in the past by other
people. As a consequence, it is still completely unclear whether the
statements in the example in 7.4.11 page 23:40-44 and 24:1-2 are
actually correct.
Indeed, in the absence of any progress guarantee, it is doubtful that
example A.2.1 is permitted to always return zero from the call to
EVENT_QUERY, thus going into an infinite loop irrespective of whether
any events had been posted. If that is allowed, it would not have a
defined interpretation, and would therefore not be a standard-conforming
program (Fortran 2008 1.5 paragraph 1). See also Atomics.2 and
General.1.
Unless those questions can be resolved, and I do not see how, I remain
of the opinion that EVENT_QUERY should be removed. If not, its effect
and whether example A.2.1 will work at all should be explicitly stated
to be processor dependent.
TEAMS
-----
Teams.1) The clarification that variables of type TEAM_TYPE have value
semantics is welcome, but the TS also needs to state whether the values
can be assigned between images and used on an image other than the one on
which they were defined. It doesn't matter whether they can or cannot,
but it should be explicitly stated which.
Teams.2) It is strange to provide intrinsics like GET_TEAM and TEAM_ID
yet provide no way to test whether two team values are the same.
Equality and inequality comparison should be supported for TEAM_TYPE.
After 5.2 page 9:30+ add:
The module defines the following:
The elemental operator == for two values of type TEAM_TYPE to
return true if the values are the same and false otherwise.
The elemental operator /= for two values of type TEAM_TYPE to
return true if the values differ and false otherwise.
Teams.3) The current wording for CHANGE TEAM still contains some
implications of association semantics, which forbids VALUE and intrinsic
copying. It has also been extended since N1996 to allow changing to the
initial team, which is not in accordance with N1981, and I assume is an
accident.
5.3 page 10:18-21 should replace:
The <team-variable> shall have been defined by execution of a FORM
TEAM statement in the team that executes the CHANGE TEAM statement
or be the value of a team variable for the initial team. The values
of the <team-variable>s on the imag
es of the team shall be those
defined by execution of the same FORM TEAM statement on all the
images of the team.
by:
The values of <team-variable> shall have been established by
execution of a FORM TEAM statement in the team that executes the
CHANGE TEAM statement. The values of the <team-variable>s on all
of the images of the team shall have been established by execution
of the same FORM TEAM statement.
Teams.5) The wording of SYNC TEAM seems to allow mixing values created
by FORM TEAM with the initial ones, and is ambiguous about exactly which
team it is collective over. It also refers to "the team variables for
the initial team", which doesn't make sense as there will usually be no
such variables.
5.6 page 12:21-25 should replace:
The SYNC TEAM statement is an image control statement. The value of
<team-variable> shall have been established by execution of a FORM
TEAM statement by the current team or an ancestor of the current
team, or be the value of a team variable for the initial team. The
values of the <team-variable>s on the images of the team shall be
those defined by execution of the same FORM TEAM statement on all
the images of the team or shall be the values of the team variables
for the initial team.
by:
The SYNC TEAM statement is an image control statement. The value of
<team-variable> shall have been established by execution of a FORM
TEAM statement by the current team or an ancestor of the current
team, or shall specify the initial team. The values of the
<team-variable>s on the images of the team specified by the
<team-variable> shall all specify the same team.
Teams.6) The example in 7.4.15 page 26:8-19 still asserts more than is
in the normative text. The whole reason for adding NEW_INDEX to FORM
TEAM was to allow control over the image ordering but leave the default
ordering processor dependent. This is not a mere nitpick - examples
like this are a major cause of erroneous interpretations by
programmers.
7.4.15 page 26:15 should be replaced by:
: ! Code for half of the images in the current team
7.4.15 page 26:17 should be replaced by:
: ! Code for the other half of the images in the current team
Teams.7) 5.5 page 12:7 refers to "the same team", but that is not
wholly clear. A Note should be sufficient. See below.
Teams.8) There is still nothing said about when resources may be
released, which is a sure recipe for some programmers assuming that they
will be and some implementors not doing it. A simple Note would be
better than nothing.
The following Note addresses both points. After 5.5 page 12:18+,
append:
NOTE
Each execution of the FORM TEAM statement establishes a new team
for each value of <team-id>. Separate executions or separate
statements establish separate teams, even if the images and values
of <team-id> are identical. Programmers should not assume that
FORM TEAM can be called an indefinite number of times, though
processors are encouraged to support that.
ATOMICS
-------
Atomics.1) 7.2 and A.3.2 are truly massive improvements and, as far as I
can see, make the intent, requirements and non-requirements clear.
However, I think that there should be a paragraph saying explicitly that
the matter of progress is not addressed by the TS, and is processor
dependent.
A.3.2.2 page 46:9+, add:
This includes the concept often called 'progress'. A change to an
atomic variable in a segment may become visible to another image
'immediately', or may not until that image reaches a segment that
is ordered after the segment in which it was changed.
Atomics.2) In 7.2 page 17:20-21, the TS says that developing a formal
data consistency model is left until integration into the main standard,
but it is important to note that such a model is NOT a matter of the
atomics alone. It will be much harder to specify the interactions
between the atomics, collectives, events and the existing execution
order model, as well as when a processor is required to make progress
(if ever). See also Event.2 and General.1.
Collectives
-----------
Collectives.1) This is an old objection, where I was mollified by
people saying that intrinsics need not follow the same rules as
ordinary procedures. However, on thinking it over, that is not an
adequate answer.
Variable definition context is strictly defined (i.e. not processor
dependent) in Fortran 2008 16.6.7 and elsewhere, and some constraints
(especially C539) require a compile-time diagnostic to be issued if a
read-only entity is used in one. The question is when SOURCE
constitutes a variable definition context. N2007 does not say so, and
it is not one of the items listed in 16.6.7.
a) If it always is, then collectives cannot be used on INTENT(IN)
arguments altogether, which is very poor software engineering.
b) If it never is, then we have defined a circumstance under which a
variable can become defined that is not a variable definition context,
which is worse.
c) It cannot be dependent on whether it can be modified, as that
is not statically determinable, though that is all that can be deduced
from the current wording.
d) It could be dependent on the presence of a RESULT argument. That
introduces a new syntactic concept into the base standard, which is
dubiously within the scope of N1981, but is feasible.
Collectives.2) The specification introduces a 'gotcha' where code like
the following will cause data corruption:
SUBROUTINE Collective (array, target_image)
INTEGER :: array, target_image
CALL CO_MAX(array,target_image)
END SUBROUTINE Collective
While that is the user's mistake, the standard prevents any kind of
diagnostic, which is poor software engineering.
Collectives.3) They are unnecessarily restrictive, in that there is no
good reason that the result should not be obtainable on an arbitrary
subset of images.
Collectives.4) While I can see why CO_BROADCAST is exceptional in having
solely an in-place form, there are circumstances under which a copying
form is wanted. I agree that a foolish consistency is the hobgoblin of
little minds, but this is a case where there are good reasons to be
consistent, even if the extended function is rarely needed.
If this specification is preserved, at least the first issue should be
resolved in normative text, and preferably the others as well.
As an example of what I regard as a better approach, here is a rewrite
of CO_MAX. The replicated wording for SOURCE and OBJECT could easily be
condensed, but I have not done so here.
7.4.7 CO_MAX (SOURCE [, RESULT, RESULT_IMAGE, STAT, ERRMSG]) or
CO_MAX_IN_PLACE (OBJECT [, RESULT_IMAGE, STAT, ERRMSG])
Description. Compute elemental maximum value on the current team of
images.
Class. Collective subroutine.
Arguments.
SOURCE shall be of type integer, real, or character. It is an
INTENT(IN) argument. It shall have the same type and type parameters on
all images of the current team. If it is a scalar, the computed value
is equal to the maximum value of SOURCE on all images of the current
team. If it is an array it shall have the same shape on all images of
the current team and each element of the computed value is equal to the
maximum value of all the corresponding elements of SOURCE on the images
of the current team.
RESULT (optional) shall be of the same type, type parameters, and shape
as SOURCE. It is an INTENT(OUT) argument.
OBJECT shall be of type integer, real, or character. It is an
INTENT(INOUT) argument. It shall have the same type and type parameters
on all images of the current team. If it is a scalar, the computed
value is equal to the maximum value of OBJECT on all images of the
current team. If it is an array it shall have the same shape on all
images of the current team and each element of the computed value is
equal to the maximum value of all the corresponding elements of OBJECT
on the images of the current team.
[[[ Unchanged
RESULT_IMAGE (optional) shall be a scalar of type integer. It is an
INTENT(IN) argument. If it is present, it shall be present on all
images of the current team, have the same value on all images of the
current team, and that value shall be the image index of an image of the
current team.
STAT (optional) shall be a scalar of type default integer. It is an
INTENT(OUT) argument.
ERRMSG (optional) shall be a scalar of type default character. It is an
INTENT(INOUT) argument.
]]]
For the purpose of matching the sequence of invocations of collective
subroutines [7.3 paragraph 1], CO_MAX and CO_MAX_IN_PLACE are the same
collective.
If RESULT_IMAGE is present, either OBJECT or RESULT shall be present on
the image specified by RESULT_IMAGE, and the computed value is assigned
to that argument. If RESULT_IMAGE is not present, the computed value is
assigned to OBJECT or RESULT on all the images of the current team that
have either argument present.
[ Issues:
1) Would it be better to say that any argument associated with
OBJECT or RESULT on other images than one specified by RESULT_IMAGE
becomes undefined? Or not used? Or processor dependent? Or simply
unspecified, as in N2007? I have no strong view.
2) If no image has either OBJECT or RESULT, this is a waste of time,
but is perfectly well-defined. That could easily be forbidden, though
I can't see why it need be.
]
[[[ Unchanged
The effect of the presence of the optional arguments STAT and ERRMSG is
described in 7.3.
Example. If the number of images in the current team is two and SOURCE
is the array [1, 5, 3] on one image and [4, 1, 6] on the other image,
the value of RESULT after executing the statement
CALL CO_MAX(SOURCE, RESULT) is [4, 5, 6] on both images.
]]]
GENERAL
-------
General.1) I remain extremely concerned about potential inconsistencies
and unimplementabilities in the specification, especially caused by
interactions between features. In particular, several aspects
introduced since N1958/N1863 are beyond the domain of established
computer science - and, as far as I know, support for 'process' failure
has never been successfully standardised before, despite several
attempts over the past half-century.
The point here is not that this TS should be delayed until it has a data
consistency model, but that we cannot be certain that one is even
definable. Half a century of experience in this area has taught us that
intuition is not a reliable guide, and 'obvious truths' are often false
when parallel data consistency is involved. As an example in a much
more widely-used language than Fortran, Java has had threading support
and a memory model since about 1995 but it was shown, only in 2004, to
be badly flawed; it was replaced in 2005.
http://www.ibm.com/developerworks/library/j-jtp02244/index.html
In the case of the facilities in this TS, I cannot either produce
examples of deadlock or causal loops, or convince myself that they do
not exist. However, I can convince myself that livelock and infinite
loops (see Events.2), in the case where the program's logic contains no
such flaws, are possible with the current specification of EVENT_QUERY.
I do not believe that developing a proper data consistency model, formal
or informal, will be possible by February 2015. As mentioned in
Atomics.2 and Events.2, this is NOT a matter of the atomics alone, but
of the interactions between the atomics, collectives, events and the
existing execution order model, as well as when a processor is required
to make progress (if ever).
A proposal is made below.
General.2) I also remain extremely concerned about whether this
specification will be reliably and efficiently implementable, for a
sufficiently high proportion of user programs, except on systems where
the hardware or operating system has special support for it. I accept
that people who understand the implementation issues will be able to
write portably efficient programs, and SOME users will do so, but
the the question is how many.
A specific issue here is that Fortran is maintaining its position
largely because it can be used to write more efficient code than its
competitors. If, in the eyes of the user community, it loses that edge,
it will disappear. In particular, I cannot see how to to implement it
portably, reliably and efficiently, using only the reliably portable
facilities of POSIX, MPI and TCP/IP. These do not include MPI passive
one-sided communication or even MPI_THREAD_MULTIPLE, largely because
there is no one-sided synchronisation in POSIX.
My understanding is that there are, as yet, only two full and released
implementations of Fortran 2008 coarrays (Cray's and Intel's), and very
little user experience with the latter; IBM's may be imminent, but I
know of no others. In all cases, the implementation works only with the
vendor's own MPI, as far as I know. It seems very unlikely indeed that
there will be significant experience with implementing and using the
facilities in this TS before February 2015.
A proposal is made below.
General.3) This comment is made mainly because it supports the previous
points in this section.
It is dubious that this TS meets requirement S1 in N1981 any longer, as
it is over double the length as and much more complex than the original
proposal (N1858 and N1863). Events can be regarded as replacing
NOTIFY/QUERY and parallel I/O has been dropped, but major new,
interacting, features include the new atomics, non-hierarchical team
usage, and support for image failure.
As a consequence of the above comments, I believe that this TS should be
explicitly removed from consideration for inclusion in the next Fortran
standard (see N1979), because the first CD ballot of the next revision
of the standard is due in February 2015.
_______________________________________________________________________
David Muxworthy
I vote abstain with two minor editorial comments:
1. Corrigendum 1 should also be listed on page 3 because Fortran
corrigenda are not cumulative. Corrigendum 3 will have almost
certainly have been approved by the time this WG5 ballot is completed
(SC22 ballot deadline today, 2014-04-15) and so should also be listed.
2. Technical Corrigendum 1 changed the heading of 13.7.165 from
THIS IMAGE ( ) or THIS IMAGE (COARRAY [, DIM])
to
THIS IMAGE ( ), THIS_IMAGE (COARRAY) or THIS_IMAGE (COARRAY, DIM)
The edits at 35:27-29, 35:32 should reflect this.
______________________________________________________________________
John Reid
2) Yes, but I recommend the following changes.
1. Failed images
There has not been adequate discussion of an image stalling because of
the failure of another image, which will cause it to fail despite there
being nothing wrong with its hardware. All we have is a warning in NOTE
5.7. If restarting is planned, as in the example A.1.2, such images
should be available for reuse. Such stalling anyway indicates that the
team calculation has gone wrong, so I suggest addding a RENDEZVOUS
specifier to the SYNC ALL statement. If a stalled image is detected, all
images of the team would re-execute the most recently executed
RENDEZVOUS FOR ALL statement with STAT= value set to STAT_FAILED_IMAGE,
ignoring all pending synchronizations initiated since that FOR ALL was
last executed. Alternatively, all images of the team could exit the
change team construct. That would give the required effect in A.1.2
without any code change.
The programmer who wishes to guard against stalling because of accessing
an image that has failed may do do as follows:
IF(ALL(FAILED_IMAGES()/=i)) THEN
a = a[i]
ELSE
! Do something else
END IF
This is illustrated in my rewrite of A.2.1 below. I think we may need
a logical function that tests a given image for failure.
2. Teams
I am unhappy with the idea a code using an image selector for ancestor
team not visible in the scope. For example, a library code should leave
such work to the calling code.
We have chosen to address an ancestor in an image selector by using its
team variable name. We should be doing the same for other references to
an ancestor. An ancestor team may be at different team distances on
executing images of different descendant teams. I suggest the
replacement of DISTANCE by TEAM as arguments of TEAM_ID, NUM_IMAGES,
and THIS_IMAGE, the removal of DISTANCE as an argument of GET_TEAM,
and the removal of TEAM_DEPTH.
3. Edits
[10:27] Change "Apart from its final upper bound, its" to "Its".
[There is nothing special about the codimension-decl here, so
there is no need to say anything about the final codimension.]
[10:30] Change "established." to "established, apart from its
final upper cobound".
[Here, the final upper cobound is likely to be different.]
[13:10-13] Replace by "The value of the default integer scalar
constant STAT_FAILED_IMAGE is positive and different from the value
of STAT_STOPPED_IMAGE, STAT_LOCKED, STAT_LOCKED_OTHER_IMAGE, or
STAT_UNLOCKED. If the processor detects that an image of the current
team has failed, the".
[With the addition of FAIL IMAGE, we want a processor that cannot
detect a truly failed image to respond to the execution of
FAIL IMAGE.]
[15:6-24] Should an event variable be atomic?
[15:34] Wording like that at [17:12-14] is needed here.
[23:26] EVENT_QUERY should be an atomic subroutine.
[40:4, 40:25-32] To make the code tolerate failing spare images,
replace the DO loop by the following:
k = images_used
DO i = 1, size(failed_img)
IF (failed_img(i) == 1) ERROR STOP 'cannot recover'
DO k = k+1, num_images()
IF (all(failed_img(:)/=k) EXIT
END DO
IF (me == k) THEN
me = failed_img(i)
id = 1
EXIT
END IF
END DO
If this is done, [40:4] should be deleted.
[43:6-44:34] This example has bugs. I think the message would
be clearer if this example were replaced by an example that is
not tolerant to failed images, followed by a modification that
is. Here is a draft.
A.2.1 EVENT_QUERY example
The following example illustrates the use of events via a program in
which image 1 acts as master and shares out work items to the other
images. Only one work item at a time can be active on a worker image,
and each deals with the result (e.g. via I/O) without directly feeding
data back to the master image.
Because the work items are not expected to be balanced, the master
keeps cycling through all the images to find one that is waiting for
work.
An event is posted by each worker to indicate that it has completed
its work item. Since the corresponding variables are needed only on
the master, we place them in an allocatable array component of a
coarray. An event on each worker is needed for the master to post the
fact that it has made a work item available for it.
PROGRAM work_share
USE, INTRINSIC :: iso_fortran_env
USE :: mod_work, ONLY: & ! Module that creates work items
work, & ! Type for holding a work item
create_work_item, & ! Function that creates work item
process_item, & ! Function that processes an item
work_done ! Logical function that returns true if all work done
TYPE(event_type) :: submit[*] ! Whether work ready for a worker
TYPE :: asymmetric_event
TYPE(event_type), ALLOCATABLE :: event(:)
END TYPE
TYPE(asymmetric_event) :: free[*] ! Whether worker is free
TYPE(work) :: work_item[*] ! Holds all the data for a work item
INTEGER :: count, i, n, nbusy[*]
IF (this_image() == 1) THEN
! Get started
ALLOCATE(free%event(2:num_images()))
nbusy = 0 ! This holds the number of workers working
DO i = 2, num_images() ! Start the workers working
IF (work_done()) EXIT
nbusy = nbusy + 1
work_item[i] = create_work_item()
EVENT POST (submit[i])
END DO
! Main work distribution loop
master : DO
image : DO i = 2, num_images()
CALL EVENT_QUERY(free%event(i), count)
IF (count == 0) CYCLE ! Worker is not free
EVENT WAIT (free%event(i));
nbusy = nbusy - 1
IF (work_done()) CYCLE
nbusy = nbusy + 1
work_item[i] = create_work_item()
EVENT POST (submit[i])
END DO image
IF ( nbusy==0 ) THEN ! All done. Exit on all images.
DO i = 2, num_images()
EVENT POST (submit[i])
END DO
EXIT master
END IF
END DO master
ELSE
! Work processing loop
worker : DO
EVENT WAIT (submit)
IF (nbusy[1] == 0) EXIT
CALL process_item(work_item)
EVENT POST (free[1]%event(this_image()))
END DO worker
END IF
END PROGRAM work_share
A.2.1a EVENT_QUERY example that tolerates image failure
This example is an adaptation of the example of A.2.1 to make it able
to execute in the presence of the failure of one or more of the worker
images. The function create_work_item now accepts an integer argument
to indicate which work item is required. It is assumed that the work
items are indexed 1, 2, ... . It is also assumed that if an image fails
while processing a work item, that work item can subsequently be
processed by another image.
The internal subroutine failed tests whether a particular image has
failed.
PROGRAM work_share
USE, INTRINSIC :: iso_fortran_env
USE :: mod_work, ONLY: & ! Module that creates work items
work, & ! Type for holding a work item
create_work_item, & ! Function that creates work item
process_item, & ! Function that processes an item
work_done ! Logical function that returns true if all work done
TYPE(event_type) :: submit[*] ! Whether work ready for a worker
TYPE :: asymmetric_event
TYPE(event_type), ALLOCATABLE :: event(:)
END TYPE
TYPE(asymmetric_event) :: free[*] ! Whether worker is free
TYPE(work) :: work_item[*] ! Holds all the data for a work item
INTEGER :: count, i, k, kk, n, nbusy[*], np, status
INTEGER, ALLOCATABLE :: working(:) ! Items being worked on
INTEGER, ALLOCATABLE :: pending(:) ! Items pending after image failure
IF (this_image() == 1) THEN
! Get started
ALLOCATE(free%event(2:num_images()))
ALLOCATE(working(2:num_images()), pending(num_images()-1))
nbusy = 0 ! This holds the number of workers working
k = 1 ! Index of next work item
np = 0 ! Number of work items in array pending
DO i = 2, num_images() ! Start the workers working
IF (work_done()) EXIT
work_item[i] = create_work_item(k)
working(i) = k
k = k + 1
nbusy = nbusy + 1
EVENT POST (submit[i], STAT=status)
IF (status==STAT_FAILED_IMAGE) THEN
working(i) = 0
k = k - 1
nbusy = nbusy - 1
END IF
END DO
! Main work distribution loop
master : DO
image : DO i = 2, num_images()
IF (ANY(FAILED_IMAGES()==i)) THEN ! Image has failed
IF (working(i)>0) THEN ! It failed while working
np = np + 1
pending(np) = working(i)
working(i) = 0
END IF
CYCLE image
END IF
CALL EVENT_QUERY(free%event(i), count)
IF (count == 0) CYCLE image ! Worker is not free
EVENT WAIT (free%event(i))
nbusy = nbusy - 1
IF (np>0) THEN
kk = pending(np)
np = np - 1
ELSE
IF (work_done()) CYCLE image
kk = k
k = k + 1
END IF
nbusy = nbusy + 1
working(i) = kk
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
END DO image
IF ( nbusy==0 ) THEN ! All done. Exit on all images.
DO i = 2, num_images()
EVENT POST (submit[i],STAT=status)
IF (status==STAT_FAILED_IMAGE) CYCLE
END DO
EXIT master
END IF
END DO master
ELSE
! Work processing loop
worker : DO
EVENT WAIT (submit)
IF (nbusy[1] == 0) EXIT worker
CALL process_item(work_item)
EVENT POST (free[1]%event(this_image()))
END DO worker
END IF
END PROGRAM work_share
_______________________________________________________________________
Van Snyder
Comments fall into three categories:
1. Stuff that might be profitably added.
2. Stuff I would do differently.
3. Other stuff
Stuff that might be profitably added
====================================
It is processor dependent whether common or independent random number
generators are used, and there's no way to detect which is the
situation. Therefore taking action using one assumption or the other
(e.g., calling or not calling RANDOM_SEED on each image with a
hopefully-different value, maybe dependent on the image number) might be
the wrong thing to do. It is preferable to provide a named constant in
ISO_FORTRAN_ENV to indicate whether the processor uses a common random
number generator on all images, or intrinsic procedures so to inquire or
specify. References to the intrinsic subroutine that specifies whether
common or independent random number generators are used should be image
control statements. 13.5p4 should be adjusted so that the segment
ordering requirement applies only if a common random number generator is
used.
Note 5.8 suggests that there ought to be a mechanism to specify, change,
and inquire the image to which standard input is connected. The OPEN
and INQUIRE statements come immediately to mind. The image index should
be in the initial team, and connecting should, for consistency, be
allowed in only one image. Either that, or standard input should be
connected to all images. A READ statement for a file that is attached
to more than one image should be specified to execute as if in a
critical section.
A pointer can have a coarray target, but cannot be directly coindexed,
although if they're actual arguments associated with (necessarily
nonpointer, so far) coarray dummy arguments, their target can be
coindexed in the invoked procedure. Allow coarrays to be pointers.
Prohibit allocating and deallocating coarray pointers. In pointer
assignment, require that if the pointer is a coarray, the data target
shall be a coarray.
Stuff I would do differently
============================
[12:6] It's a bit easier to compute team numbers if zero is allowed.
E.g., mod(this_image,4). Replace "be greater than zero" with "not be
negative" and replace "1" with "zero" at [26:6]
[15:14-15] It would be better if variables of EVENT_TYPE were not
required to be coarrays. Instead eliminate C601 and constrain
<event-variable> to be a coarray, a coindexed object, or a pointer in
C604, and a coarray or a pointer in C605. In pointer assignment, if a
<data-pointer-object> is of type EVENT_TYPE, or of a type that has a
noncoarray potential subobject component of type EVENT_TYPE, constrain
the <data-target> to be a coarray or a pointer. Thereby, by induction,
a EVENT_TYPE pointer can only be associated with a coarray. 14-138
advocates the same for LOCK_TYPE.
[15:16-24 C602, C603] Protected types would be a better solution.
If EVENT_TYPE were a protected type, or by adding
"C602a A subobject of an event variable shall not appear in a variable
definition context, shall not be an actual argument in a
reference to a procedure with implicit interface, and shall not
be an actual argument associated with a dummy argument that has
unspecified intent."
at [15:24+], it would be possible to expose the event count as a
specific component name, and eliminate EVENT_QUERY.
Since we have decided that we do not want to allow type-bound procedures
to be actual arguments or procedure pointer targets, we could allow to
invoke type-bound functions that have no arguments without parentheses
enclosing the (empty) argument list. A corresponding change should
allow to reference a scalar with an empty subscript list. Thereby, it
would be impossible to know, and impossible for the syntax to depend
upon, whether event%count or event%count() is a reference to a scalar
component or a type-bound function that has no arguments. This is
intended as food for future thought; it's outside the scope of the
present TS.
If protected types existed, LOCK_TYPE should also be a protected type.
[17:24-25] Since references to collective subroutines have to be invoked
by the same statement on all nonfailed images of the current team, and
it's hard to imagine how they can produce an output argument value
that's the result of a collective action until all of them finish, it
would be simpler either to make references to them be image control
statements (and establish segment boundaries), or to specify that
executing them includes the effect of executing an image control
statement. (This is done by the edit proposed for [32:21+] below.)
This would also affect the discussion at [47:15-16].
[18:11ff General for 7.4.2, 7.4.4 and 7.4.5] Since intrinsic procedures
are "magic" it would be possible (and more economical) to replace
ATOMIC_AND, ATOMIC_OR and ATOMIC_XOR with ATOMIC_OP ( ATOM, OPERATOR,
VALUE ) or ATOMIC_OP ( ATOM, OPERATOR, VALUE, OLD ), and require
OPERATOR to be a local name for one of the intrinsic functions IAND,
IEOR or IOR. Yes, this is an exception to the general rule that generic
names cannot be actual arguments, but so what? Intrinsics are magic.
[18:30ff General for 7.4.7, 7.4.8] Since intrinsic procedures are
"magic" it would be possible (and more economical) to delete CO_MAX and
CO_MIN in favor of CO_REDUCE, and allow the OPERATOR of CO_REDUCE to be
a local name for one of the intrinsic functions MAX or MIN. It would be
nice to include PRODUCT and SUM under this rubric, but they don't have
two arguments, so explanation of CO_REDUCE might become difficult. Yes,
this is an exception to the general rule that generic names cannot be
actual arguments, but so what? Intrinsics are magic.
[24:20-25:20] Since TEAM_TYPE is not prohibited in a variable
definition context, GET_TEAM could be a function instead of a
subroutine. This would be handy for passing team variables as actual
arguments.
[32:21+] Insert "{In 8.5.1 Image control statements, insert a list item}
'a CALL statement that references an intrinsic collective subroutine'".
Other stuff
===========
[General] Turn on paragraph numbering.
[General] Process with LaTeX such that ligatures are not used. This
makes words that contain, e.g., "fi", searchable in Acrobat Reader.
[General] Use either "in the current team" or "of the current team", and
either "in the initial team" or "of the initial team", throughout, not
one or the other at random. Using "in" might make more sense since
"image of the current team" need not necessarily be read as "image that
is a member of the current team." It might be read as if there is,
somehow, an "image" (with a meaning different from the one in Subclause
2.3.4 of 1539-1, e.g., "a photograph", or a copy -- "image" -- of the
team variable), that reflects or represents all of the current team.
[General] "positive" appears in some places, while "greater than zero"
appears in other places. Use "positive" throughout (because it's
shorter).
[General] "less than or equal to" appears in some places, while "not
greater than" appears in other places. Use "not greater than"
throughout (because it's shorter).
[Foreword iii:p5:line 3] Is "must" the correct word here?
[Introduction iv:p1:2] Delete "a set of".
[Introduction iv:p2:6] Delete "for".
[Introduction iv:p2:8] Replace "progress" with "proceed" because
"progress" does not appear anywhere else, except in notes, in either the
TS or 14-007. Either that, or define "progress" (as a verb) in the
"Terms and definitions" subclause.
[Introduction iv:p3:5] Delete "sets of".
[9:26-30] Should TEAM_TYPE be specified to have a public type-bound
defined assignment? Or does it not matter because one cannot tell the
difference?
[10:7 C503] Replace "A" with "An".
[10:13 C505] After "<coselector-name>" insert ", or the same as a
<coarray-name> in another <codimension-decl>,".
[10:31-34] Raises the question whether allocatable non-coarray variables
that are not allocated at the beginning of a CHANGE TEAM construct are
deallocated at the end. Probably not. If yes, it should be specified.
If not, it should be mentioned in a note.
[10:36] The phrase "team containing the executing image that is
identified by <team-variable>" should almost certainly be "current
team". I don't see how things can be otherwise. If some subset of the
current team decide they're going to be in one subteam, and only those
images execute the CHANGE TEAM statement, from the point of view of
other images of the (previously) current team, the images of that
subteam are still part of their team; the result is overlapping teams,
until other images, that are members of other teams, execute a different
CHANGE TEAM statement, maybe in an unordered segment. I thought the
intent of the CHANGE TEAM construct was to partition the current team,
so that no image can ever be a member of more than one
currently-executing team. In any case, the phrase "executing image that
is identified by <team-variable>" appears to be nonsense. A
<team-variable> identifies a set of images, not one image. I don't see
a clean way to reorganize the sentence to make it clear that
"<team-variable> applies to "the team", not "the executing image".
[12:6-7] Replace "be greater than zero and is the same for all images
that are members of the same team" with "be positive. Images are
members of the same team if and only if they specify the same value of
<team-id>." (I prefer "not be negative" instead of "be positive". It's
a bit easier to compute team numbers if zero is allowed. See the remark
for [12:6] in the "Stuff I would do differently" section above.)
[12:12] Replace "assigned by the processor" with "a processor dependent
value that shall be positive and not greater than the number of images
in the team".
[13:Note 5.6] There are assumptions about RANDOM_NUMBER here. If all
images produce exactly the same sequence of random numbers, they will
all fail, pretty much at the same time. Maybe insert something to the
effect "if the random number generators on different images are
independent and are started with seeds that produce different sequences,
or a common random number generator is used...."
[13:20] Replace "defined" with "declared".
[14:Note 5.8] Replace "image 1" with "image 1 in the initial team"
twice.
[15:7] Append "It does not have the BIND attribute, and is not a
sequence type."
[15:9-10] Replace "The effect of each change ... change" with "The event
count is updated as if by invoking the ATOMIC_ADD intrinsic subroutine"
to avoid any suspicion that a new kind of mechanism is in play here.
[15:16] Replace "An" with "A nonpointer".
[15:19,24] Replace "where" with "if" twice.
[15:21] Insert "nonpointer" before "variable".
{It would be better to append "unless both the actual argument and dummy
argument are pointers" at the end of 16.6.7p1(12), but that ought to be
done in the normal course of a revision, not sneaked in via a TS.}
[15:24+] Insert a constraint
"C603a (R425) If EXTENDS appears and the type defined has a potential
subobject component of type EVENT_TYPE from the intrinsic module
ISO_FORTRAN_ENV, its parent type either shall be EVENT_TYPE or
shall have a potential subobject component of type EVENT_TYPE."
This is actually undesirable to put in the standard. What we really
want to do is to combine the protection of EVENT_TYPE and LOCK_TYPE from
by polymorphism, so that one could extend EVENT_TYPE and give a
component of LOCK_TYPE and vice-versa; therefore to re-write C435 to say
C435 (R425) If EXTENDS appears and the type defined has a potential
subobject component of type EVENT_TYPE or LOCK_TYPE, from the
intrinsic module ISO_FORTRAN_ENV, its parent type either shall be
EVENT_TYPE or LOCK_TYPE, or shall have a potential subobject
component of type EVENT_TYPE or LOCK_TYPE.
[15:24+] Insert a paragraph
"The value of the actual argument corresponding to the CPTR argument of
the C_F_POINTER subroutine from the ISO_C_BINDING intrinsic module shall
not be the address of an object of type EVENT_TYPE, unless the FPTR
argument is a pointer of the same type as the object."
[15:24+] Note 13.29 in 14-007, concerning lock variables, was apparently
considered to be useful. Insert a similar note concerning event
variables:
"Note 6.0
The restrictions against changing an event variable except via
EVENT POST and EVENT WAIT statements ensure the integrity of its
value and facilitate efficient implementation, particularly when
special synchronization is needed for correct event handling."
[15:26, 30-31] Replace the introductory paragraph with
"Successful execution of an EVENT POST statement posts an event,
consisting of atomically incrementing the count of the event variable
by 1. If an error condition occurs during execution of an EVENT POST
statement, the count does not change. The EVENT POST statement is an
image control statement."
Then delete the paragraph immediately after C604.
Using "atomically" here is important. Either that, or it is necessary
to say that an EVENT POST statement executes as if in a critical
section.
[15:29] Delete "the".
[16:1-16] Notwithstanding "the effect of each change is as if it
occurred instantaneously" at [15:8-9], it seems to be necessary to say
that steps (2) and (3) of the EVENT WAIT statement execute as if in a
critical section, because testing its value isn't part of "each change
... occurred instantaneously." Otherwise, if two images execute EVENT
WAIT statements, when another image posts an event, it is possible that
both of them notice the threshold is exceeded, both stop waiting, and both
reduce the event count, the result being that the event count becomes
negative. Indeed, it might be necessary to say this about some other
image control statements, at least ALLOCATE, DEALLOCATE, and calls to
MOVE_ALLOC. If so, can it be done here, or is an interp needed?
[16:2] Replace the introductory paragraph with
"Successful execution of an EVENT WAIT statement waits until a
specified number of events are posted by successful execution of
EVENT POST statements. The EVENT WAIT statement is an image control
statement."
[16:8] Replace "threshold of its event argument" with "threshold value".
Otherwise, "threshold value" in the next two steps is not a defined
concept. Don't use "argument" since the EVENT WAIT statement is not a
procedure reference.
[16:12] Replace "event count" with "count of the event variable", for
consistency with [15:30] and [16:10].
[17:32] Delete first "the".
[17:33] Replace "on" with "in".
[17:38] Insert "processor-dependent" before "nonzero". Should there be
a requirement that the nonzero value be different from
STAT_STOPPED_IMAGE or STAT_FAILED_IMAGE?
[18:1] Insert "the named constant" before "STAT_STOPPED_IMAGE" and
replace "in" with "from".
[18:3] Insert "the named constant" before "STAT_FAILED_IMAGE" and
replace "in" with "from".
[18:4] Replace "as" with "to be" or "to have".
[18:10+ Note 7.1] I can't figure out the point of the note. If it isn't
pointless, replace "undefined ... collective with RESULT" with
"undefined if an error condition occurs during execution of a collective
subroutine and the RESULT argument is".
[18:16 19:2] Delete "and" twice.
[19:9-11] Can we put the argument value conditions first? For example
"CALL ATOMIC_AND ( I[3], 6, IOLD ) with the value of I[3] equal to 5
causes...."
[19:16 19:34] Delete "and" twice.
[19:41-43] Can we put the argument value conditions first?
[20:5] Delete "and".
[20:1 20:12 34:14+9] Perhaps ATOMIC_XOR should be ATOMIC_EOR since the
intrinsic function it says it uses "under the covers" is IEOR.
[20:7] Repair the typesetting so the line doesn't end with left
parenthesis.
[20:12-14] Can we put the argument value conditions first?
[20:19-22] The SOURCE argument to CO_REDUCE is prohibited to be
polymorphic. Can the SOURCE argument to CO_BROADCAST be polymorphic? If
so, insert "dynamic" before "type" and replace "parameters" with
"parameter values" (the parameters are necessarily the same on all
images, because all images execute the same statement). If not, replace
the sentence with "shall be a scalar or array of any type. It shall not
be polymorphic. Each length parameter shall have the same value on all
images of the current team." The first paragraph of 7.3 requires the
subroutine to be invoked by the same statement on all images of the
current team, and therefore the declared type and kind type parameter
values cannot be different.
[19:22] Append "It is an INTENT(INOUT) argument." The associated actual
argument cannot be an expression because all images execute the same
statement, and a new value is assigned on all but SOURCE_IMAGE. This
cannot be done for CO_MAX etc., for reasons already explained by Nick.
[20:34-35] Delete "It shall have... team." because it cannot be
otherwise.
[20:40,41] Either replace "RESULT" with "it" at line 40, or replace "It"
with "RESULT_IMAGE" at line 41.
[21:5] Delete "the" before "images".
[21:7, 21:34, 23:15] Is it really necessary for SOURCE to become
undefined on images other than RESULT_IMAGE if RESULT is not present?
Why not simply "unchanged"?
[21:20-21] Delete "It shall have... team." because it cannot be
otherwise.
[21:26-27] Either replace "RESULT" with "it" at line 26, or replace "It"
with "RESULT_IMAGE" at line 27.
[21:32] Delete "the" before "images".
[22:1-2] ER-RMSG is strange hyphenation. If it doesn't fit if
hyphenated ERR-MSG, put it all on the next line.
[22:6-7] Delete "It shall have... team." because it cannot be
otherwise.
[22:17,19] Either replace "RESULT" with "it" at line 17, or replace "It"
with "RESULT_IMAGE" at line 19.
[22:44-23:1] Delete "It shall have... team." because it cannot be
otherwise.
[23:7-8] Either replace "RESULT" with "it" at line 7, or replace "It"
with "RESULT_IMAGE" at line 8.
[23:13] Delete "the" before "images".
[23:24-24:1] Since EVENT_QUERY almost certainly internally consists
only of looking at a component of a coindexed object, why does it have
STAT and ERRMSG arguments? A reference to a coindexed object in, say,
an expression, doesn't have them. It would be simpler if EVENT_QUERY
were a function, maybe better as a type-bound function than an intrinsic
function. Even better would be if EVENT_TYPE were a protected type and
COUNT were a public component.
[23:30] Replace "no" with "not".
[23:32] Subtracting one for each successfully-executed EVENT WAIT
statement doesn't account for the UNTIL_COUNT specifier in the EVENT
WAIT statement. But why specify the computation here (again)? Replace
"number of successful posts minus the number of successful waits" with
"the event count of the EVENT argument".
[23:35] Replace "defined" with "dependent".
[25:36] Is it possible for EVENT_QUERY to set STAT to STAT_STOPPED_IMAGE
or STAT_FAILED_IMAGE? Even if not, should there be a requirement that
the error status be different from those values (at least to
future-proof it)?
[24:10] Insert "decimal" before "range". Replace "be at least as large
as" with "not be less than that" (for consistency with [23:30]).
[26:6] Replace "1" with "zero" if it is decided that team numbers can
begin at zero (see the edit above for [12:6-7] and the remark for [12:6]
in the "Stuff I would do differently" section above).
[26:40] Insert "processor-dependent" before "nonzero". Should there be
a requirement that the nonzero value be different from
STAT_STOPPED_IMAGE or STAT_FAILED_IMAGE (at least to future-proof it)?
[27:12,14] After "present" insert "or is present with the value false",
then delete ", otherwise ... specified".
[29:9] Replace "a" with "as".
[29:16] Replace "it" with "the subset" to avoid confusion with "it"
meaning "the program".
[29:21] Insert "for" before "performing".
[29:23] Replace "features for the support of" with "facilities to
support".
[29:24] Replace "features" with "facilities".
[30:11] After "team" insert ", consisting of all images,"
[31:2] Is it necessary or useful to mention access to images of other
teams here?
[31:13] Insert a comma after "appears".
[33:35-32:5] Are these the only possible nonzero values? If not, add a
sentence about any other errors resulting in a processor-dependent
positive value different from STAT_....
[31:37] Replace "in" with "from" for consistency with 6.7.4 etc.
[32:1] Replace "in" with "from".
[32:4] Replace "in" with "from".
[32:5] Insert "allocation" before "status".
[32:36] Delete "that is".
[33:17-25] Are these the only possible nonzero values? If not, add a
sentence about any other errors resulting in a processor-dependent
positive value different from STAT_....
[33:20,21,24] Replace "in the intrinsic" with "from the intrinsic"
thrice.
[33:37+] Insert a new subclause
"8.7a Edits to Clause 9
{In subclause 9.5.1, paragraph 4, concerning an asterisk in a READ
statement, insert 'in the initial team' after 'image 1'.}"
[34:7+] Insert "{In Note 13.1 replace 'atomic variables' with 'access to
variables using atomic subroutines'}" because Fortran does not have
atomic variables, and the term is not used anywhere else.
[34:19+] Insert "{In subclause 13.5, paragraph 3, insert 'in the initial
team' after 'image 1'}." [In a separate J3 paper, I advocate that the
kinds of specific discussions in 13.5p3-6 ought to be in the individual
procedure descriptions, not in 13.5.]
[34:19+] Insert "{In subclause 13.5, paragraph 4, replace 'the
interleaving of calls to RANDOM_NUMBER' with 'the sequences of values
assigned to the HARVEST argument to RANDOM_NUMBER'}." because the
interleaving of calls to all subroutines in unordered segments is
processor dependent (although a standard-comforming program cannot
observe that), so what's currently in 13.5p4 doesn't say much of
anything.
[35:24-26] After "present" insert "or is present with the value false",
then delete ", otherwise ... specified".
[36:6-7] Replace "and" with a comma. At the end of the sentence append
", and the final normative paragraph" [the one added above by the edit
at [15:24+].
[36:7+] Insert "{In 13.8.2.16 LOCK_TYPE, in constraints C1303 and
C1404, replace 'where' with 'if' twice.}".
[36:9] If "5.6" is not generated by \ref, replace "5.6" with "5.8". If
it is, check the \ref, and either repair it or run LaTeX a few more
times.
[36:9+] Subclause 5.8 requires the value of STAT_FAILED_IMAGE to be
different from the values of STAT_STOPPED_IMAGE, STAT_LOCKED,
STAT_LOCKED_OTHER_IMAGE and STAT_LOCKED, but there are no requirements
on the values of the latter constants other than the existing
requirement that the value of STAT_STOPPED_IMAGE be different from the
value of IOSTAT_INQUIRE_INTERNAL_UNIT. Should that be corrected here,
or in the normal course of revising the base standard? If the latter,
the TS would need to anticipate the revision and amend it. Therefore,
rather than putting the requirements from subclause 5.8 of the TS in a
new subclause 13.8.2.21b, append a new subclause 13.8.26 "Uniqueness of
values of named constants" with text something like "The values of the
named constants IOSTAT_INQUIRE_INTERNAL_UNIT, STAT_FAILED_IMAGE,
STAT_LOCKED, STAT_LOCKED_OTHER_IMAGE, STAT_STOPPED_IMAGE, and
STAT_UNLOCKED shall be distinct."
[36:11+] Insert "{In C1303 in 13.8.2.16, insert 'in an ALLOCATE
statement without a SOURCE= specifier, as an <allocate-object> in a
DEALLOCATE statement' after '<allocate-object>'.}"
Insert "{In C1304 in 13.8.2.16, insert 'in an ALLOCATE statement without
a SOURCE= specifier, as an <allocate-object> in a DEALLOCATE statement,'
after '<allocate-object>'.}"
Insert "{In 13.8.2.16 insert a paragraph
'The value of the actual argument associated with the CPTR argument of
the C_F_POINTER subroutine from the ISO_C_BINDING intrinsic module shall
not be the address of a variable of type LOCK_TYPE, unless the FPTR
argument is a pointer of the same type as the object.'
}"
[36:11+] Insert a subclause 8.9a Edits to clause 15:
"{Insert the final normative paragraph of 6.2 at the end of the
description of the FPTR argument to C_F_POINTER in subclause 15.2.3.3.}"
[36:6-7] Replace "and" with a comma. At the end of the sentence insert
"and constraint C604b. Move C604b to be immediately after C435 in
subclause 4.5.2.1".
[36:11+] Insert "{In 13.8.2.16 LOCK_TYPE, insert 'nonpointer' before
'variable' in C1303 and C1304.}"
[36:12+] Insert "{In 16.4 Statement and construct entities, in paragraph
1, after 'DO CONCURRENT' replace 'or' with a comma; after 'ASSOCIATE
construct' insert ', or as a coarray specified by a <codimension-decl>
in a CHANGE TEAM construct,'}".
[36:14-15] The characteristics of associating entities of the CHANGE
TEAM construct are described in 5.3. Replace "and bounds... coselector"
with ", bounds, and cobounds specified in [whatever new subclause number
of 14-007 subclause 5.3 of the TS becomes]."
[36:18+] Insert "{After item (13) in 16.6.7, insert a list item}
'(13a) a <coarray> in a <codimension-decl> in a CHANGE TEAM construct
if the coarray named by the corresponding <coselector-name> of
that construct [", or a subobject thereof," if 14-136 passes]
appears in a variable definition context within that
construct;'"
[36:23+] Insert "{In the list item concerning COMMAND_ARGUMENT_COUNT
etc., insert 'in the initial team' after 'image 1'}".
[36:23+] Insert "{In the second list item concerning subclause 13.7,
replace 'CMDSTAT or STATUS' with 'CMDSTAT, STAT, or STATUS'.}"
Alternatively, change the STAT argument to collective subroutines,
EVENT_QUERY, and MOVE_ALLOC, to STATUS, and change nothing at
[14-007:468:42]. See also the edits suggested for [36:31+] below, which
ought not to be done if the alternative is adopted.
[36:25] Include an instruction to insert subclause numbers in the
newly-inserted list items.
[36:30] Replace with "the relative order of execution of EVENT POST
statements in unordered segments;" for consistency with [15:33-34].
[But isn't the relative order of everything in unordered segments
processor dependent?]
[36:31+] If the STAT argument of collective subroutines, EVENT_QUERY,
and MOVE_ALLOC is not changed to STATUS, add three more processor
dependencies:
" o the value of the STAT argument to collective subroutines if an error
occurs (13.???);
o the value of the STAT argument to the EVENT_QUERY subroutine if an
error occurs (13.???);
o the value of the STAT argument to the MOVE_ALLOC subroutine if an
error occurs (13.???);"
[40:3] Insert "in the initial team" after "image 1".
[47:11] Delete "in".
_______________________________________________________________________
More information about the J3
mailing list