-
Notifications
You must be signed in to change notification settings - Fork 7
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
Terms Chapter RC Review #335
Comments
need[s?] not -> is not required to |
As far as I understood from discussions in the web, "MPI need not deallocate ..." is correct because "need" is used as a modal verb and therefore (at least in Amercan English) does not get the "s" and does not get the "to". "it [does?] not use" seems to really require the "does". In this late state, I would always prefere to only correct the grammer and not to substitute the wording. |
Added some additional comments from @abouteiller |
A list of all MPI-4.0 procedures in the Function index and #characters of each function name, see |
Good catch on 2.5.4 - the wrong macros were used for the second list. That one is fixed in PR https://github.com/mpi-forum/mpi-standard/pull/345 |
Of these in Section 2.5.8,
Line 13 is fixed in PR 350 through the addition of a \ctypeshort macro. Line 20 is awkward, but is intentional - the intent was to keep the "INTEGER (KIND=xxx)" together, and that led to the bad break. |
The names problem is even harder: In some new constants and callback prototypes and one existing MPI-3.1 callback prototype, the 25 year old limit of 30 characters for any MPI names is violated: |
Is this issue now fully addressed or are there outstanding points to be dealt with later? |
15:30 above: This break is because the next token does not fit. This is working as intended. To remove/reduce the break, a re-write of the paragraph would be required. |
I Handled the Reviewer comments in PR 440. The two not checked I rejected as unneeded or incorrect (is there any way to not that on the checkboxes?) |
@wgropp Thanks a lot, seems we are (almost) done with this! I rearranged the checkboxes to mark some points raised as REJECTED. Concerning editor changes not done and therefore still unchecked, I assume these are also rejected (twice & fine with me) or you forgot to check (once; last point). If you agree with my evaluation please check the 3 checkboxes. |
Yes, the unchecked ones were rejected. That last one was corrected by changing the macro; I'm not sure that the sizes are the same, but the macros selected are now correct. |
Problem
These are comments on the RC from @bgoglin, @abouteiller and @RolfRabenseifner:
From @bgoglin:
Section 2.5.1
needs?
Section 2.5.4
The interline space between constants seems bigger in the first list than in the second list.
Section 2.5.8
Not sure why counts and typemaps are plural while others are not.
There are some linebreaks that don't seem properly placed, causing the line to look too short:
Section 2.9.2
line 6:
it does not use?
Additional comments from @abouteiller:
pdf 50, chap 2.2, pp 10:9
Need to verify if statement “MPI identifiers are limited to 30 characters)” is still valid. We may have spilled over with the large count variants and some session procedure names.
pdf 52, chap2.4.1, pp12:40
Sentence “Additionally, an MPI operation can be collective or noncollective” starts a new paragraph with a major change of topic. The formatting and indentations are not conducive of that shift. The sentence appears to be part of the prior bullet list item.
To a lesser extent, pp12:8 has the same problem.
Comments from @RolfRabenseifner:
Chap2.4.1, pp12:8-9, 40, 45-56
It is already mentioned by @abouteiller, but this problem is on more locations and should be done totally with noindent,
because some of these heading lines are only one line and some are more than one line.
Chap 2.2, pp 9:40
In C and the Fortran mpi_f08 module, all ...
Chap 2.2, pp 9:42
In the Fortran mpi module and mpif.h file, ...
Chap 2.2, pp 9:46-47
If the routine is not associated with a class, the name should be of the form
MPI_Action_subset or MPI_ACTION_SUBSET in C and Fortran.
Chap 2.2, pp 10:9
As already mentioned by @abouteiller, we have a problem with the 30 character procedure name limit
(only routines without MPI_T_...):
. . . . .
(comment was moved to #403)
Todo
References
All changes and their formatting are checked and verified in
https://github.com/mpi-forum/mpi-issues/files/5725671/mpi40-report_Issue405_PR439.pdf
The text was updated successfully, but these errors were encountered: