[Mpi-forum] Chapter Reviews

Mon Jul 23 06:02:14 CDT 2012

Alexander,

thank you for your detailed review on Annex 1.

Details:
-----------------
> 665/14:	Remove comma in "C, and Fortran". Apparently, C++ was there. :)
done.
> 665/16:	Replace "chapter" by "section" (tentative).
not done.
> 665/22:	Replace " is listed in the left column" by "s are listed below". The table has only one column now. C++, once again.
done.
> 665/22:	Probably, the whole first sentence should go: there are many Fortran-only tables below, and much more.
I let it there.
> 665/25 (table):	Where is MPI_ERR_NOT_IMPLEMENTED or something? Would be good for those parts of the MPI-3 standard that will naturally lag behind others or cannot be provided on the target platform (like those new windows).

Partial implementation / subsetting is not supported by the MPI standard. One may use MPI_OTHER.

> 667/4-17:	Do not use abbreviations; add underscores to the MPI_T error code names, like: MPI_T_CANNOT_INIT, and so on (otherwise difficult to read and unlike old error codes, but for the MPI_ERR_LASTCODE). If accepted, also affects the MPI_T sections.

I'll forward this to the tools group.
They changed CANTINIT into CANNOTINIT. They keep the names without underscores.
Underscores are used only for the structural parts.

It is currently under discussion and will be fixed this week.

> 669/12:	Add "signed " ahead of "long long" (like in the previous line).
Done as on page 26
> 669/25:	Why bother pointing out the header? Remove this line.
It is as on page 26

> 671/1:	Remove " and C++" in the parentheses. No C++ binding, right?
Yes, removed ecverywhere.

> 674/43:	Add "." (period) at the end of the sentence.
done.

> 676/29-30:	Fix indentation.
Done.

> 676/36:	Fix indentation.
Done.

> 677/29 (table):	Some, if not all, of these are predefined memory locations in Fortran. Would be good to highlight this, like in the earlier table w/ the MPI_BOTTOM inside.
Done.

> 680/9-10:	Add underscores for better legibility. If accepted, impacts the MPI_T chapter.
Not done -  same reason as for MPI_T_ERR... names.

> 681/1:	"and mpi module" - could've been handled by a special Fortran 2008 compatibility module instead.
> 681/4:	Same here.

Decided long time ago in the language chapter. Can't be reported other here in the appLang chapter.

> 682/20:	Here and elsewhere: (optionally) add after this line
>   END SUBROTINE
> END INTERFACE

We always show only the important header lines and never the trailing lines.
Therefore  would not change this.
p684:36 - p686:37 and same lines in MPI-2.2 also do not 
have the END SUBROUTINE lines.

> 682/31:	Same.
> 682/37:	Same.
> 682/48:	Same.

> 683/6:	Same.
> 683/16:	Same.
> 683/23:	Same.
> 683/29:	Same.
> 683/36: Same.
> 683/43:	Same.

> 684/5:	Same.
> 684/10:	Same.
> 684/16: Same.
> 684/25:	Same.
> 684/34:	Same.

Dito.

> 759/25-27:	Remove these lines: empty sections.

It is hard to do because they are generated automatically.
I'll ask Bill.

> 788/18:	Remove this line: the section is empty.

It is hard to do because they are generated automatically.
I'll ask Bill.

Thanks again for your detailed and helpful review.

Best regards
Rolf

----- Original Message -----
> From: "Alexander Supalov" <alexander.supalov at intel.com>
> To: "Main MPI Forum mailing list" <mpi-forum at lists.mpi-forum.org>
> Sent: Friday, July 20, 2012 7:28:54 AM
> Subject: Re: [Mpi-forum] Chapter Reviews
> Done (see attached). What is the procedure to vet and implement these
> changes?
> 
>  
> 
> By the way, whoever is going to review chapter 17, please look into
> 602/1-6:
> 
>  
> 
> Compliant MPI-3 implementations providing a Fortran interface must
> provide provide
> 
> one or both of the following:
> 
> _ The USE mpi_f08 Fortran support method.
> 
> _ The USE mpi and INCLUDE 'mpif.h' Fortran support methods.
> 
>  
> 
> Double “provide” aside, this passage amounts to opening the door to a
> silent removal (not even deprecation) of the old Fortran interface.
> This cannot be the real intent of the Forum, I hope. Also, some
> advices to the implementors effectively suggest breaking backward
> compatibility on the old Fortran interface, if retained, up to and
> including a failed application build and changes to the application
> necessary as a consequence. Again, I think this is a bit too much of a
> good thing.
> 
>  
> 
> 
> 
> From: mpi-forum-bounces at lists.mpi-forum.org
> [mailto:mpi-forum-bounces at lists.mpi-forum.org] On Behalf Of Richard
> Graham
> Sent: Thursday, July 19, 2012 5:29 PM
> To: Main MPI Forum mailing list
> Subject: Re: [Mpi-forum] Chapter Reviews
> 
>  
> 
> Feel free to take it – thanks !!!!
> 
>  
> 
> Rich
> 
>  
> 
> 
> 
> From: mpi-forum-bounces at lists.mpi-forum.org
> [mailto:mpi-forum-bounces at lists.mpi-forum.org] On Behalf Of Supalov,
> Alexander
> Sent: Thursday, July 19, 2012 11:02 AM
> To: Main MPI Forum mailing list
> Subject: Re: [Mpi-forum] Chapter Reviews
> 
>  
> 
> Hi Rich,
> 
>  
> 
> It might help those interested if you described the applicable
> deadlines, the desired output of the review, and the resp. correction
> procedure, should it become necessary.
> 
>  
> 
> I hate to see my favorite Appendix A orphaned. Anyone against my
> taking on it?
> 
>  
> 
> Best regards.
> 
>  
> 
> Alexander
> 
>  
> 
> 
> 
> From: mpi-forum-bounces at lists.mpi-forum.org
> [mailto:mpi-forum-bounces at lists.mpi-forum.org] On Behalf Of Richard
> Graham
> Sent: Thursday, July 19, 2012 4:45 PM
> To: mpi-forum at lists.mpi-forum.org
> Subject: [Mpi-forum] Chapter Reviews
> 
>  
> 
> We are currently doing chapter reviews for the forum – all tickets are
> committed and reviewed.  If anyone has changes to make please
> coordinate this morning (Thursday 7/19/2012 – Chicago time) with the
> person who is doing the review.  After these are done, responsibility
> will revert to the chapter committees.  I will send out a separate
> note on the timelines we agreed to yesterday, but in a nutshell, all
> changes must be completed and reviewed by Thursday 7/26/2012, 5pm
> Pacific time, so that we can prepare a draft version to be posted for
> comment.
> 
>  
> 
> If someone remote would like to take the responsibility for an
> unclaimed chapter, please let me know.
> 
>  
> 
> Thanks,
> 
> Rich
> 
>  
> 
> Chapter               Reviewer
> 
> Front Matter   Mohamad
> 
> Chap 1                Brian B.
> 
> Chap 2                Scott
> 
> Chap 3                 Anh
> 
> Chap 4        Terry
> 
> Chap 5        Nathan
> 
> Chap 6        Darius
> 
> Chap 7        Shinji
> 
> Chap 8        Pavan
> 
> Chap 9        Howard   - done
> 
> Chap 10      Hideyuki
> 
> Chap 11      Howard
> 
> Chap 12      Brian S.
> 
> Chap 13      Brian S.
> 
> Chap 14      Rich
> 
> Chap 15
> 
> Chap 16     
> 
> Chap 17
> 
> Appendix A
> 
> Appendix B     Jeff/Rolf
> 
> Bib
> 
> Indices
> 
> Intel GmbH
> Dornacher Strasse 1
> 85622 Feldkirchen/Muenchen, Deutschland
> Sitz der Gesellschaft: Feldkirchen bei Muenchen
> Geschaeftsfuehrer: Douglas Lusk, Peter Gleissner, Hannes Schwaderer,
> Christian Lamprechter
> Registergericht: Muenchen HRB 47456
> Ust.-IdNr./VAT Registration No.: DE129385895
> Citibank Frankfurt a.M. (BLZ 502 109 00) 600119052 Intel GmbH
> Dornacher Strasse 1
> 85622 Feldkirchen/Muenchen, Deutschland
> Sitz der Gesellschaft: Feldkirchen bei Muenchen
> Geschaeftsfuehrer: Douglas Lusk, Peter Gleissner, Hannes Schwaderer,
> Christian Lamprechter
> Registergericht: Muenchen HRB 47456
> Ust.-IdNr./VAT Registration No.: DE129385895
> Citibank Frankfurt a.M. (BLZ 502 109 00) 600119052
> _______________________________________________
> mpi-forum mailing list
> mpi-forum at lists.mpi-forum.org
> http://lists.mpi-forum.org/mailman/listinfo.cgi/mpi-forum

-- 
Dr. Rolf Rabenseifner . . . . . . . . . .. email rabenseifner at hlrs.de
High Performance Computing Center (HLRS) . phone ++49(0)711/685-65530
University of Stuttgart . . . . . . . . .. fax ++49(0)711 / 685-65832
Head of Dpmt Parallel Computing . . . www.hlrs.de/people/rabenseifner
Nobelstr. 19, D-70550 Stuttgart, Germany . (Office: Allmandring 30)