Rajeev
/************* I/O Chapter *************/
210/1: No comma before "and"
(because a rule in writing says that if "and" is used to connect two
clauses of a sentence, no comma should be used before "and"
if the subject of the second clause lies in the first clause.
e.g., I did this and then did that. (no comma)
I did this, and then I did that. (comma)
)
210/4: The words "These displacements" were fine in the previous
version, because the previous sentence refered to displacements. Now
that the previous sentence has changed, "These displacements" should
be replaced by "The displacements in the typemap of the filetype"
210/16: "during program execution"
211/47: No comma before "and"
211/48: "Furthermore, the use of" instead of "Further, use of"
212/2: no comma before "and"
212/38: OR should be in capitals
214/35: no comma before "and"
215/22: "that" instead of "which"
215/22-23: A better sentence:
The values of data in the new regions in the file (those
locations with displacements between old file size and \mpiarg{size})
are undefined.
215/36: "that" instead of "which"
215/38: should be plural "sizes"
216/6: A better sentence:
\func{MPI\_FILE\_PREALLOCATE} ensures that storage space is
allocated for the first \mpiarg{size} bytes of the file
associated with \mpiarg{fh}.
219/42: no comma after "group"
Also, "These target nodes coalesce" instead of "These coalesce"
219/44: Which additional hints? They should be listed. I suggest the
foll. sentence instead:
Collective buffering parameters are further directed via
the additional hints: \info{cb\_block\_size},
\info{cb\_buffer\_size}, and \info{cb\_nodes}.
219/47: "distributed among" instead of "distributed to"
220/2: comma instead of semi-colon after "node"
220/26: A better sentence:
Setting this hint is only useful when passed to \func{MPI\_FILE\_OPEN}
with an \mpiarg{amode} that includes \const{MPI\_MODE\_CREATE}.
220/44-45: Split the sentence into two.
period after "devices"
followed by: It is expressed in bytes.
221/29: no comma before "and"
221/42: delete "the" after "follow"
222/15.5: "by using" instead of "using"
222/39: no comma after "null info"
223/20: comma before "respectively"
223/18-26: Are the etype and filetype returned
by this function committed or noncommitted? A clarifying
statement is needed.
Note that the etype and filetype passed to MPI_File_set_view are
required to be committed.
223/33-35: Why are the curly braces "}" there on these three lines
after "vs."? In Latex, you should use vs.\ to prevent extra space
after the period.
224/21: no comma after "operations"
224/25: Add "For writes" as follows:
For writes, however, the \func{MPI\_FILE\_SYNC} routine ...
224/39: "that" instead of "which"
226/4: Before "The layout of data ..." add this important sentence:
The \mpiarg{datatype} passed to the routine must be
a committed datatype.
226/24: "by using" instead of "using"
235/16: Use "To set" instead of "In order to set"
240/30: Use "To set" instead of "In order to set"
240/37: no comma after "data accesses"
241/2: no comma after "open"
241/5: no comma after "open"
246/26: The increased interoperability outside the environment is also
due to tbe external data representation, not just the data conversion
functions. Therefore, I suggest changing the last part of the sentence
to:
and supports increased interoperability outside that environment
through the external data representation (Section ?, page ?) as
well as the data conversion functions (Section ?, page ?)
(Fill in the labels for section and page nos.)
246/30: add comma after "process" and add comma after "page255)"
246/32: "Furthermore" instead of "Further"
246/47: "Furthermore" instead of "Further"
247/21-22: "on each file open" is not accurate, because the data
representation is specified in file_view. Use the foll. sentence instead:
Therefore, to correctly retrieve file data, an \MPI/
application is responsible for specifying the same data
representation as was used to create the file.
247/27: no comma after "conversions"
use "within a purely ..." instead of "in a purely ..."
248/7: Give a reference to where the "data conversion rules for
communication" are. They are in Section 3.3.2, pg. 25-27, of the MPI-1
document.
249/32: comma after "formats"
249/33: comma after "112 fraction bits"
249/34: add "in" between "are" and "two's"
251/3: comma after MPI_TYPE_CREATE_F90_COMPLEX
251/11: "User-Defined" (add hyphen)
251/18: "user-defined"
252/29-30: This sentence should be:
\MPI/ will only call this routine with predefined datatypes
employed by the user.
("employed by the user" is important here)
253/5: Add this sentence after "native representation."
\mpiarg{datatype} will be equivalent to the datatype that the user
passed to the read or write function.
(This sentence is specified for the write_conversion_fn on the
next page, 254/11, but should be specified for the
read_conversion_fn as well. )
254/29: "Furthermore" instead of "Further"
254/30: no comma after "routines"
254/38: "that" instead of "which"
254/43: Since there no "two systems" discussed in the previous
sentences, delete "on both systems"
The meaning is clear even without it.
255/1-8: The three conditions essential for compatibility are listed
here. Since the second condition is required only for Fortran 90
programs, it should say so. i.e., the second bullet should begin
with:
In the case of Fortran 90 programs, the programs participating ....
255/9: "User-defined" (hyphen)
255/12: "that" instead of "which"
255/18: Add "either" before "nonconcurrent" to make it clear
255/26: "Case 3" instead of "case 3"
257/8: semi-colon instead of comma after "set"
257/36: semi-colon instead of comma after "enabled"
258/48: There isn't a section 4.13 in the MPI-1.1 document.
The correct section is Section 4.12.
259/1-2: This advice to users no longer makes sense, because the
sentence before it and one critical word in it have both been
deleted in the final edits. I suggest you delete the advice
altogether. As it currently stands, it is wrong. For
multiple collective operations to the same file, you don't need to
open the file multiple times. You just open it once and issue the
collective calls one after another on the same file handle.
259/12-14: The part beginning with "the current etype must..." is not
precise enough. The precise version is on 226/7.
Therefore, use the foll. sentences instead:
In general, the etype of data items written must match the etype
used to read the items. For each data access operation,
the type signature of the datatype used in the read or write call
must match the type signature of some number of contiguous copies
of the current etype.
259/36: Use "as values of type" instead of "by an"
259/45: "\MPI/" sounds better, instead of "This standard"
260/20: "file sizes" (plural)
262/41: "The above program also violates" instead of
"This program violates"
262/41: no comma after "operations"
264/47: "that" instead of "which"
265/30: Instead of "open" use \mpifunc{MPI\_FILE\_OPEN}
/***************** Miscellany Chapter *************/
72/5: Since this datatype constructor is not just for constructing
filetypes, I suggest the foll. sentence instead:
The \mpiarg{order} argument specifies the storage order for the
subarray as well as the full array.
74/41: for the same reason as above, just delete the last part of this
sentence:
"for multidimensional arrays in memory and in the file"
/***************** External Interfaces *************/
169/21: comma after "windows"
170/2: "operations" (plural)
170/5: no comma after "system"
171/45: "user-allocated" (hyphen)
172/15: typo: should be \mpifunc{free\_fn}
172/18: comma after "point"
181/48: comma after "way"
182/2: comma after "used"
182/11: "user-defined" (hyphen)
182/12-13: The foll. sentence reads better:
The value returned by this key is always greater than or equal
to \consti{MPI\_ERR\_LASTCODE}.
182/33: no comma after errorclass
183/15-16: Since this sentence is not related to MPI_Add_error_string,
there should be a blank line preceding it.
185/44: an error that a spell-checker will not catch:
"latter" instead of "later"
186/27: comma after "Fortran"
add "used" after "and the latter is"
186/30: comma after "call"
187/46: "returned datatypes" (plural)
187/48: no comma after "predefined datatype"
207/32-33: "and yields" instead of "as well as yielding"
/************* For the whole document ************/
In general, when something in quotes is followed by a comma or a
period, the comma or period must be included in the quotes.
i.e., ``foo,'' instead of ``foo'',
(It doesn't seem logical, but it's a rule in writing.)
This rule has been correctly followed in some places in the document,
but in some places it is not.
grep \'\'\, *.tex
and
grep \'\'\\. *.tex
should catch the errors.