10.1.1
P205-L31: "are monotonically non-decreasing. Data access"
P205-L41: "to be distinct, but they cannot be negative must be
monotonically non-decreasing."
10.2.1
P208-L9: "to end of file." (period added)
P208-L11: "C/C++ users" (lowercase "u")
I would add a sentence to explain what MPI_DELETE_ON_CLOSE stands for,
before P208-L27.
I would move the sentence on P208-L25-26 to P208-L37 (just before the
Advice to users).
I would move the paragraph on P208-L41-43 to P208-L37 (just after the
sentence just moved).
I would move the sentence on P208-L45-46 and the Advice to Implementors on
P209-L1-5
to P207-L32 (just before the Advice to users).
10.2.3
P210-L11: "error handler assigned to MPI_FILE_HANDLE_X (see section 10.8)."
10.2.4
P210-L25: I would add a reference to section 10.6.7 for the definition of
MPI_OFFSET_KIND or
I would redefine it here
P210-L28-29: "the file handle fh. MPI_FILE_SET_SIZE is collective;" The
sentence I suggest to
delete is redundant with the next paragraph
P210-L36: "use MPI_PREALLOCATE to force" (delete "below")
10.2.7
P213-L12: "code fragment"
10.2.8
P213-L27: "MPI_OPEN" and "MPI_FILE_SET_VIEW" are not the only routines
which specify hints.
"MPI_FILE_SET_INFO" and "MPI_FILE_DELETE" do too
P215-L46: "should be used to store the file"
P216-L5-6: I cannot recall the initial definition of num_io_nodes. I
thought it was the number of I/O devices
that should be used to store the file (when io_node_list
is not specified and when all I/O devices
are kind of equivalent for the application). This number
could be smaller than the number of I/O
devices in the system. And I think we should also
specify: "This hint is most relevant when the
file is created."
P216-L11: "the amount of consecutive data assigned to one I/O"
10.3
P218-L6 is stronger than the sentence on P240-L36-37. I think they should
agree, and if P218-L6 is to be
kept, I would like to add that it is erroneous to use
MPI_TYPE_HINDEXED or MPI_TYPE_HVECTOR
when constructing an etype or filetype. In addition,
sentences on P205-L29-31 and on P205-L37-38
may be respectively incorrect and ambiguous if P218-L6 is to
be kept
10.4
P219: In the table, the two-phase collective data access routines are
missing
10.4.3
P225-L10,48: "by MPI" or "by the MPI implementation"
P226-L34: "by MPI" or "by the MPI implementation"
P227-L7: "by MPI" or "by the MPI implementation"
P227-L34: A figure showing what the end of the view represents would help
P227-L48: "relative to the current view of fh"
10.4.4
P228-L45-46: This sentence is not true for SHARED_ORDERED routines
P231-L10-11: "within the file handle communicator group."
P231-L18: "file handle communicator group need to access"
P231-L48/P232-L1: The sentence "This routine ... file handle fh." is
superfluous.
It just defines again what a collective I/O
operation is.
P232-L7-8: "MPI_READ_SHARED_ORDERED stores the amount of data read in
status.
The number of etype or datatype elements read can be
extracted from status
using the routines MPI_GET_COUNT or MPI_GET_ELEMENTS."
P232-L30-31: The sentence "All processes ... must call
MPI_WRITE_SHARED_ORDERED"
is also superfluous.
P232-L7-8: "MPI_WRITE_SHARED_ORDERED stores the amount of data written in
status.
The number of etype or datatype elements written can be
extracted from status
using the routines MPI_GET_COUNT or MPI_GET_ELEMENTS."
P233-L44: "to the current view of fh"
10.4.5
P234-L16: "here is intended to make"
P234-L23: "When an ``end'' call"
P234-L32: "an MPI_Read_all and an MPI_Read_all_start/end"
P235-L17: "and MPI_READ_ALL_END) compose"
P235-L27: "DATATYPE, IERROR"
P235-L41: to be deleted
P236-L3: "DATATYPE, IERROR"
P236-L16: to be deleted
P236-L28: "DATATYPE, IERROR"
P236-L39: to be deleted
P237-L3: "DATATYPE, IERROR"
P237-L15: to be deleted
P237-L26: "DATATYPE, IERROR"
P237-L38: to be deleted
P238-L2: "DATATYPE, IERROR"
P238-L14: to be deleted
10.5
P240-L5: "``external32''"
P240-L7: "``external32''"
P240-L9: didn't we decide to call the canonical representation ``external''
instead of
``external32'' since there is only one now (if this is the
case all occurrences
of ``external32'' should be replaced with ``external''.
10.5.2
P242-L37: "newtype is a new datatype"
10.5.3
In addition to Steve Huss-Lederman's comments:
P242-L12: "passed to the MPI_REGISTER_FILTER call"
P242-L39: "passed to the MPI_REGISTER_FILTER call"
P246-L1-2: "for a user-defined data representation. If a matching"
(From John May's revised version - formatting his Tex source file)
P2-L30: "the first converted datum at the location"
P3-L10: "continue the storing of converted data" or "continue storing
converted data"
P3-L13: "cache an internal representation"
P3-L34: "with the same datatype and userbuf arguments and appropriate
values"
10.5.4
P242-L19: "``external32'' representation" if ``external32'' remains the
name for the
canonical representation
10.6.6
P250-L17: extra space at the end of the line that can be avoided
10.6.7
P250-L28: extra space at the end of the line that can be avoided
10.7.1
P251-L6-9: The sentence "Note that these datatype constructors ... in this
section."
is not relevant any longer since MPI_TYPE_SUBARRAY and
MPI_TYPE_DARRAY
are now fundamental datatypes
10.7.2
P251-L43-44: line 44 should be a continuation line (MPI macro problem)
P253-L8: "integers)"
P253-L11: "(array of integers)"
P253-L13-14: "(array of positive integers)"
P253-L17: "integers)"
P253-L29-30: line 30 should be a continuation line (MPI macro problem)
P253-L40-43: Since argument pdims has been removed, this paragraph must be
modified
accordingly (pdims should be replaced by ndims)
P254-L32-34: Since argument pdims has been removed, these lines should now
become:
"array_of_psizes(1) = 2
array_of_psizes(2) = 1
array_of_psizes(3) = 3"
10.10.1
P257-L35: "Example local array filetype for process 1"
Jean-Pierre