June 2011 - mpi-forum - mailman3.cels.anl.gov

Draft for FT Run-Through Stabilization available / Ticket #276
by Josh Hursey 30 Jun '11

30 Jun '11

The first reading of the Run-Through Stabilization Fault Tolerance proposal will occur during the July 2011 MPI Forum meeting. In preparation for this reading the proposal has been attached to the ticket: https://svn.mpi-forum.org/trac/mpi-forum-web/ticket/276 The text is also available in the SVN in the 'ft' MPI-3 working group section. The ticket changes in the PDF are mostly marked by 'ticket276' but also include the addition of Chapter 17 (which is not marked due to latex issues). Comments and feedback are always welcome. Josh -- Joshua Hursey Postdoctoral Research Associate Oak Ridge National Laboratory http://users.nccs.gov/~jjhursey

1 0

Re: [Mpi-forum] \MPIreplace and verbatim blocks
by Rolf Rabenseifner 26 Jun '11

26 Jun '11

In the Fortran chapter , I have the same problem and I used several methods, depending on the needed visibility on the Contents pages. Only changed from subsubsection to subsection: \MPIupdate{3.0}{230-B}{% \subsection{Problems Due to Strong Typing} \label{sec:f90-problems:strong-typing} }% \MPIdelete{3.0}{230-B}{% \subsubsection{Problems Due to Strong Typing} }% Changed from subsubsection to subsection and added something to the title: \subsection{Problems Due to Data Copying and Sequence Association \MPIupdate{3.0}{236-H}{with Subscript Triplets}} \label{sec:misc-sequence} \MPIdelete{3.0}{230-B}{% \subsubsection{Problems Due to Data Copying and Sequence Association} }% New subsection where I wanted that it is not only marked in the section's pages, but also onb the Contents pages: \subsection{\MPIupdate{3.0}{236-H}{Problems Due to Data Copying and Sequence Association with Vector Subscripts}} It is a rare problem, and I believe that this works good enough. The body of these subsections are marked independently. Rolf ----- Original Message ----- > From: "William Gropp" <wgropp(a)illinois.edu> > To: "Main MPI Forum mailing list" <mpi-forum(a)lists.mpi-forum.org> > Sent: Monday, June 13, 2011 11:57:00 PM > Subject: Re: [Mpi-forum] \MPIreplace and verbatim blocks > This is an example of why I would have preferred for MPI-3 to be a new > document. There is really no way to do this, because it is changing > the basic structure of the document (adding sections or even chapters > is ok, changing levels is not). The change macros can only be used to > mark changes to the text. > > What I could provide is a way to allow you to describe the old state, > rather than running the commands from the prior version. However, I > don't see much value in that. Instead, I'd use the approach in the > RMA chapter for the moved examples - simply describe what has changed > and leave it at that. > > Bill > > On Jun 13, 2011, at 4:03 PM, Martin Schulz wrote: > > > Hi Bill, > > > > On Jun 8, 2011, at 7:00 AM, William Gropp wrote: > > > >> I need to add \MPIreplaceBegin and MPIreplaceEnd for this - they > >> will > >> parallel the use of MPIupdateBegin/MPIupdateEnd. Until I do, you > >> can > >> use the MPIdeleteBegin/End and MPIupdateBegin/End . In general, the > >> Begin/End versions need to be used whenever any non-trivial TeX > >> magic > >> (such as verbatim) is taking place. > > > > Is there any way to handle situations in which section or > > subsection commands get changed? In particular, for MPI_T > > we need to change \chapter into \section, \section into > > \subsection, etc. and I am not sure how to annotate that > > so it conforms with the rest of the documents. Just using the > > begin/end macros doesn't work since it still executes both > > commands in this case. > > > > Any help on this would be great, > > Thanks, > > > > Martin > > > > > >> > >> Bill > >> > >> On Jun 8, 2011, at 8:03 AM, Jeff Squyres wrote: > >> > >>> Bill -- > >>> > >>> Is there a correct way to use \MPIreplace with {verbatim} blocks? > >>> This form seems to generate latex errors: > >>> > >>> \MPIreplace{3.0}{278}{ > >>> \begin{verbatim} > >>> old busted foo > >>> \end{verbatim} > >>> }{ > >>> \begin{verbatim} > >>> new hotness foo > >>> \end{verbatim} > >>> } > >>> > >>> (same seems to be true with {Verbatim} blocks) > >>> > >>> -- > >>> Jeff Squyres > >>> jsquyres(a)cisco.com > >>> For corporate legal information go to: > >>> http://www.cisco.com/web/about/doing_business/legal/cri/ > >>> > >>> > >>> _______________________________________________ > >>> mpi-forum mailing list > >>> mpi-forum(a)lists.mpi-forum.org > >>> http://lists.mpi-forum.org/mailman/listinfo.cgi/mpi-forum > >> > >> William Gropp > >> Director, Parallel Computing Institute > >> Deputy Director for Research > >> Institute for Advanced Computing Applications and Technologies > >> Paul and Cynthia Saylor Professor of Computer Science > >> University of Illinois Urbana-Champaign > >> > >> > >> > >> _______________________________________________ > >> mpi-forum mailing list > >> mpi-forum(a)lists.mpi-forum.org > >> http://lists.mpi-forum.org/mailman/listinfo.cgi/mpi-forum > > > > ________________________________________________________________________ > > Martin Schulz, schulzm(a)llnl.gov, http://people.llnl.gov/schulzm > > CASC @ Lawrence Livermore National Laboratory, Livermore, USA > > > > > > > > _______________________________________________ > > mpi-forum mailing list > > mpi-forum(a)lists.mpi-forum.org > > http://lists.mpi-forum.org/mailman/listinfo.cgi/mpi-forum > > William Gropp > Director, Parallel Computing Institute > Deputy Director for Research > Institute for Advanced Computing Applications and Technologies > Paul and Cynthia Saylor Professor of Computer Science > University of Illinois Urbana-Champaign > > > > _______________________________________________ > mpi-forum mailing list > mpi-forum(a)lists.mpi-forum.org > http://lists.mpi-forum.org/mailman/listinfo.cgi/mpi-forum -- Dr. Rolf Rabenseifner . . . . . . . . . .. email rabenseifner(a)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)

1 0

Hotel for the July meeting
by Graham, Richard L. 24 Jun '11

24 Jun '11

Just a reminder that Monday, 6/27/2011, is the last day to reserve a room at the Residence Inn in Chicago. Rich

1 0

ticket process clarification
by Fab Tillier 23 Jun '11

23 Jun '11

Hi Folks, Torsten recently pointed me to the standardization process document for MPI-3.0, available here: http://meetings.mpi-forum.org/standardization_process_3.pdf In the 'process for adding new functionality to the standard' section of the document: - there is no requirement for change-log entries in individual tickets, only for the final chapter review. - there is no requirement for a 2 week notice for PDF files to be available before a formal reading. There is a 1 week notice requirement for the final chapter review. - there is no definition for requirements for voting other than passing a formal reading. Do chapter authors need to sign off? Do chapter committees need to sign off? Or has the Forum as a whole signed off by the mere fact that the change passed its formal reading? - there is no definition for how soon before the voting process an implementation, if required, should be available. Is an implementation required before the 1st vote, or only before the 2nd? In the 'final chapter vote' requirements: - How can changes that have not yet been approved be part of the final chapter vote? (second bullet) It would be great to clarify these issues in the document, so as to make ticket authoring less painful than it currently is. Thanks, -Fab

1 0

Re: [Mpi-forum] \MPIreplace and verbatim blocks
by William Gropp 13 Jun '11

13 Jun '11

This is an example of why I would have preferred for MPI-3 to be a new document. There is really no way to do this, because it is changing the basic structure of the document (adding sections or even chapters is ok, changing levels is not). The change macros can only be used to mark changes to the text. What I could provide is a way to allow you to describe the old state, rather than running the commands from the prior version. However, I don't see much value in that. Instead, I'd use the approach in the RMA chapter for the moved examples - simply describe what has changed and leave it at that. Bill On Jun 13, 2011, at 4:03 PM, Martin Schulz wrote: > Hi Bill, > > On Jun 8, 2011, at 7:00 AM, William Gropp wrote: > >> I need to add \MPIreplaceBegin and MPIreplaceEnd for this - they will >> parallel the use of MPIupdateBegin/MPIupdateEnd. Until I do, you can >> use the MPIdeleteBegin/End and MPIupdateBegin/End . In general, the >> Begin/End versions need to be used whenever any non-trivial TeX magic >> (such as verbatim) is taking place. > > Is there any way to handle situations in which section or > subsection commands get changed? In particular, for MPI_T > we need to change \chapter into \section, \section into > \subsection, etc. and I am not sure how to annotate that > so it conforms with the rest of the documents. Just using the > begin/end macros doesn't work since it still executes both > commands in this case. > > Any help on this would be great, > Thanks, > > Martin > > >> >> Bill >> >> On Jun 8, 2011, at 8:03 AM, Jeff Squyres wrote: >> >>> Bill -- >>> >>> Is there a correct way to use \MPIreplace with {verbatim} blocks? >>> This form seems to generate latex errors: >>> >>> \MPIreplace{3.0}{278}{ >>> \begin{verbatim} >>> old busted foo >>> \end{verbatim} >>> }{ >>> \begin{verbatim} >>> new hotness foo >>> \end{verbatim} >>> } >>> >>> (same seems to be true with {Verbatim} blocks) >>> >>> -- >>> Jeff Squyres >>> jsquyres(a)cisco.com >>> For corporate legal information go to: >>> http://www.cisco.com/web/about/doing_business/legal/cri/ >>> >>> >>> _______________________________________________ >>> mpi-forum mailing list >>> mpi-forum(a)lists.mpi-forum.org >>> http://lists.mpi-forum.org/mailman/listinfo.cgi/mpi-forum >> >> William Gropp >> Director, Parallel Computing Institute >> Deputy Director for Research >> Institute for Advanced Computing Applications and Technologies >> Paul and Cynthia Saylor Professor of Computer Science >> University of Illinois Urbana-Champaign >> >> >> >> _______________________________________________ >> mpi-forum mailing list >> mpi-forum(a)lists.mpi-forum.org >> http://lists.mpi-forum.org/mailman/listinfo.cgi/mpi-forum > > ________________________________________________________________________ > Martin Schulz, schulzm(a)llnl.gov, http://people.llnl.gov/schulzm > CASC @ Lawrence Livermore National Laboratory, Livermore, USA > > > > _______________________________________________ > mpi-forum mailing list > mpi-forum(a)lists.mpi-forum.org > http://lists.mpi-forum.org/mailman/listinfo.cgi/mpi-forum William Gropp Director, Parallel Computing Institute Deputy Director for Research Institute for Advanced Computing Applications and Technologies Paul and Cynthia Saylor Professor of Computer Science University of Illinois Urbana-Champaign

1 0

Re: [Mpi-forum] \MPIreplace and verbatim blocks
by Martin Schulz 13 Jun '11

13 Jun '11

Hi Bill, On Jun 8, 2011, at 7:00 AM, William Gropp wrote: > I need to add \MPIreplaceBegin and MPIreplaceEnd for this - they will > parallel the use of MPIupdateBegin/MPIupdateEnd. Until I do, you can > use the MPIdeleteBegin/End and MPIupdateBegin/End . In general, the > Begin/End versions need to be used whenever any non-trivial TeX magic > (such as verbatim) is taking place. Is there any way to handle situations in which section or subsection commands get changed? In particular, for MPI_T we need to change \chapter into \section, \section into \subsection, etc. and I am not sure how to annotate that so it conforms with the rest of the documents. Just using the begin/end macros doesn't work since it still executes both commands in this case. Any help on this would be great, Thanks, Martin > > Bill > > On Jun 8, 2011, at 8:03 AM, Jeff Squyres wrote: > >> Bill -- >> >> Is there a correct way to use \MPIreplace with {verbatim} blocks? >> This form seems to generate latex errors: >> >> \MPIreplace{3.0}{278}{ >> \begin{verbatim} >> old busted foo >> \end{verbatim} >> }{ >> \begin{verbatim} >> new hotness foo >> \end{verbatim} >> } >> >> (same seems to be true with {Verbatim} blocks) >> >> -- >> Jeff Squyres >> jsquyres(a)cisco.com >> For corporate legal information go to: >> http://www.cisco.com/web/about/doing_business/legal/cri/ >> >> >> _______________________________________________ >> mpi-forum mailing list >> mpi-forum(a)lists.mpi-forum.org >> http://lists.mpi-forum.org/mailman/listinfo.cgi/mpi-forum > > William Gropp > Director, Parallel Computing Institute > Deputy Director for Research > Institute for Advanced Computing Applications and Technologies > Paul and Cynthia Saylor Professor of Computer Science > University of Illinois Urbana-Champaign > > > > _______________________________________________ > mpi-forum mailing list > mpi-forum(a)lists.mpi-forum.org > http://lists.mpi-forum.org/mailman/listinfo.cgi/mpi-forum ________________________________________________________________________ Martin Schulz, schulzm(a)llnl.gov, http://people.llnl.gov/schulzm CASC @ Lawrence Livermore National Laboratory, Livermore, USA

1 0

Re: [Mpi-forum] 2012 tentative meeting agenda
by Rajeev Thakur 10 Jun '11

10 Jun '11

Yes, June 25-28 will conflict with the SC12 PC mtg Rajeev On Jun 8, 2011, at 1:12 PM, Bronis R. de Supinski wrote: > > Rich: > > My 2012 clalendar is not fully populated yet so > it is hard to say for certain although I do not > see any conflicts for most of the dates. However, > June 25-28 is bad. It conflicts with ICS and most > probably with the SC PC meeting (its dates will > be finalized tomorrow). In general June 2012 is > already full with significant events each week. > I suggest we push that meeting out to mid to late > July 2012. > > Bronis > > > On Wed, 8 Jun 2011, Graham, Richard L. wrote: > >> At the last meeting we had general agreement to continue the current pace of meetings until MPI 3.0 is finalized, and then go down to 4 meetings a year, but longer meetings. The time frame we have set for MPI 3.0 means that the first 3 meetings in 2012 will continue at the current pace, and then we will cut back to quarterly. Here is a first suggestion for meeting dates ï¿½ please send any conflicts that you are aware of. Also, there has been a push to move the west coast meetings to Portland, OR ï¿½ are people OK with that ? >> >> Thanks, >> Rich >> >> Jan 9-11: Portland, OR >> March 5-7: Chicago, IL >> April 30 ï¿½ May 2: Portland, OR ï¿½ Final MPI 3.0 vote >> June 25-28: Chicago, IL >> Sept: In conjunction with EuroMPI >> Dec 3-6: Portland, OR > _______________________________________________ > mpi-forum mailing list > mpi-forum(a)lists.mpi-forum.org > http://lists.mpi-forum.org/mailman/listinfo.cgi/mpi-forum

1 0

Re: [Mpi-forum] (Continued) Latex woes
by William Gropp 10 Jun '11

10 Jun '11

Thanks. I've committed the fix in 278. Bill On Jun 10, 2011, at 9:41 AM, Jeff Squyres wrote: > Ah, gotcha. I just added you to the "wg" SVN group, so you should > be able to commit in the wg tree now. > > > > On Jun 10, 2011, at 10:32 AM, William Gropp wrote: > >> I've committed it in one place in the approved branch, but can't >> fix it in the bw-compat ticket 278 branch. >> >> Bill >> >> On Jun 10, 2011, at 7:08 AM, Jeff Squyres wrote: >> >>> Thanks! >>> >>> Is there an SVN permissions problem such that you can't commit? >>> (assumedly to the /minutiae branch, since the MPIreplace* (etc.) >>> macros are not on the /approved branch...?) >>> >>> >>> On Jun 9, 2011, at 8:23 PM, William Gropp wrote: >>> >>>> Attached is the diff of the fix. I can't commit it, so one of >>>> you will need to. >>>> >>>> Bill >>>> >>>> <fix> >>>> On Jun 9, 2011, at 12:20 PM, William Gropp wrote: >>>> >>>>> Floats lost is often due to a limitation in LaTeX. One possible >>>>> issue is that the the change markers use floats, and so do >>>>> tables. LaTeX isn't very happy with floats within floats. I'm >>>>> in a meeting today and can't look at it until later, but one >>>>> solution may be to inline the ticket info in these cases. >>>>> >>>>> Bill >>>>> >>>>> On Jun 9, 2011, at 11:30 AM, Jeff Squyres wrote: >>>>> >>>>>> Bill -- >>>>>> >>>>>> We're still having latex woes in ticket 278. Can you help? >>>>>> >>>>>> We replaced all the \MPIreplace macros with deletebegin/end and >>>>>> updatebegin/end, but we're still getting "Float(s) lost". It >>>>>> seems to have to do with a table in datatype.tex. The table >>>>>> looks like this: >>>>>> >>>>>> ----- >>>>>> \begin{table}[ht] >>>>>> \begin{center} >>>>>> \begin{constlist} >>>>>> \constitem{MPI\_COMBINER\_NAMED}{a named predefined datatype} >>>>>> ...etc. >>>>>> \MPIdeleteBegin{3.0}{278} >>>>>> \constitem{MPI\_COMBINER\_HVECTOR\_INTEGER}{\mpifunc{MPI\_TYPE >>>>>> \_HVECTOR} from Fortran} >>>>>> \constitem{MPI\_COMBINER\_HVECTOR}{\mpifunc{MPI\_TYPE\_HVECTOR} >>>>>> from C or C++} >>>>>> \constskipitem{}{\ \ and in some case Fortran} >>>>>> \constskipitem{}{\ \ or \mpifunc{MPI\_TYPE\_CREATE\_HVECTOR}} >>>>>> \MPIdeleteEnd{3.0} >>>>>> \MPIupdateBegin{3.0}{278} >>>>>> \constitem{MPI\_COMBINER\_HVECTOR}{\mpifunc{MPI\_TYPE\_CREATE >>>>>> \_HVECTOR}} >>>>>> \MPIupdateEnd{3.0} >>>>>> ...etc. >>>>>> ----- >>>>>> >>>>>> If we remove all the \MPIdelete* and \MPIupdate* macros, it >>>>>> builds ok. If we put even one of them back, we get "Float(s) >>>>>> lost" and pdflatex aborts. But other than that, we're totally >>>>>> stumped (we've wasted 2+ hours trying to solve this so far :-( ). >>>>>> >>>>>> What should we do? >>>>>> >>>>>> -- >>>>>> Jeff Squyres >>>>>> jsquyres(a)cisco.com >>>>>> For corporate legal information go to: >>>>>> http://www.cisco.com/web/about/doing_business/legal/cri/ >>>>>> >>>>>> >>>>>> _______________________________________________ >>>>>> mpi-forum mailing list >>>>>> mpi-forum(a)lists.mpi-forum.org >>>>>> http://lists.mpi-forum.org/mailman/listinfo.cgi/mpi-forum >>>>> >>>>> William Gropp >>>>> Director, Parallel Computing Institute >>>>> Deputy Director for Research >>>>> Institute for Advanced Computing Applications and Technologies >>>>> Paul and Cynthia Saylor Professor of Computer Science >>>>> University of Illinois Urbana-Champaign >>>>> >>>>> >>>>> >>>>> _______________________________________________ >>>>> mpi-forum mailing list >>>>> mpi-forum(a)lists.mpi-forum.org >>>>> http://lists.mpi-forum.org/mailman/listinfo.cgi/mpi-forum >>>> >>>> William Gropp >>>> Director, Parallel Computing Institute >>>> Deputy Director for Research >>>> Institute for Advanced Computing Applications and Technologies >>>> Paul and Cynthia Saylor Professor of Computer Science >>>> University of Illinois Urbana-Champaign >>>> >>>> >>>> >>>> _______________________________________________ >>>> mpi-forum mailing list >>>> mpi-forum(a)lists.mpi-forum.org >>>> http://lists.mpi-forum.org/mailman/listinfo.cgi/mpi-forum >>> >>> >>> -- >>> Jeff Squyres >>> jsquyres(a)cisco.com >>> For corporate legal information go to: >>> http://www.cisco.com/web/about/doing_business/legal/cri/ >>> >>> >>> _______________________________________________ >>> mpi-forum mailing list >>> mpi-forum(a)lists.mpi-forum.org >>> http://lists.mpi-forum.org/mailman/listinfo.cgi/mpi-forum >> >> William Gropp >> Director, Parallel Computing Institute >> Deputy Director for Research >> Institute for Advanced Computing Applications and Technologies >> Paul and Cynthia Saylor Professor of Computer Science >> University of Illinois Urbana-Champaign >> >> >> >> _______________________________________________ >> mpi-forum mailing list >> mpi-forum(a)lists.mpi-forum.org >> http://lists.mpi-forum.org/mailman/listinfo.cgi/mpi-forum > > > -- > Jeff Squyres > jsquyres(a)cisco.com > For corporate legal information go to: > http://www.cisco.com/web/about/doing_business/legal/cri/ > > > _______________________________________________ > mpi-forum mailing list > mpi-forum(a)lists.mpi-forum.org > http://lists.mpi-forum.org/mailman/listinfo.cgi/mpi-forum William Gropp Director, Parallel Computing Institute Deputy Director for Research Institute for Advanced Computing Applications and Technologies Paul and Cynthia Saylor Professor of Computer Science University of Illinois Urbana-Champaign

1 0

Re: [Mpi-forum] (Continued) Latex woes
by Jeff Squyres 10 Jun '11

10 Jun '11

Ah, gotcha. I just added you to the "wg" SVN group, so you should be able to commit in the wg tree now. On Jun 10, 2011, at 10:32 AM, William Gropp wrote: > I've committed it in one place in the approved branch, but can't fix it in the bw-compat ticket 278 branch. > > Bill > > On Jun 10, 2011, at 7:08 AM, Jeff Squyres wrote: > >> Thanks! >> >> Is there an SVN permissions problem such that you can't commit? (assumedly to the /minutiae branch, since the MPIreplace* (etc.) macros are not on the /approved branch...?) >> >> >> On Jun 9, 2011, at 8:23 PM, William Gropp wrote: >> >>> Attached is the diff of the fix. I can't commit it, so one of you will need to. >>> >>> Bill >>> >>> <fix> >>> On Jun 9, 2011, at 12:20 PM, William Gropp wrote: >>> >>>> Floats lost is often due to a limitation in LaTeX. One possible issue is that the the change markers use floats, and so do tables. LaTeX isn't very happy with floats within floats. I'm in a meeting today and can't look at it until later, but one solution may be to inline the ticket info in these cases. >>>> >>>> Bill >>>> >>>> On Jun 9, 2011, at 11:30 AM, Jeff Squyres wrote: >>>> >>>>> Bill -- >>>>> >>>>> We're still having latex woes in ticket 278. Can you help? >>>>> >>>>> We replaced all the \MPIreplace macros with deletebegin/end and updatebegin/end, but we're still getting "Float(s) lost". It seems to have to do with a table in datatype.tex. The table looks like this: >>>>> >>>>> ----- >>>>> \begin{table}[ht] >>>>> \begin{center} >>>>> \begin{constlist} >>>>> \constitem{MPI\_COMBINER\_NAMED}{a named predefined datatype} >>>>> ...etc. >>>>> \MPIdeleteBegin{3.0}{278} >>>>> \constitem{MPI\_COMBINER\_HVECTOR\_INTEGER}{\mpifunc{MPI\_TYPE\_HVECTOR} from Fortran} >>>>> \constitem{MPI\_COMBINER\_HVECTOR}{\mpifunc{MPI\_TYPE\_HVECTOR} from C or C++} >>>>> \constskipitem{}{\ \ and in some case Fortran} >>>>> \constskipitem{}{\ \ or \mpifunc{MPI\_TYPE\_CREATE\_HVECTOR}} >>>>> \MPIdeleteEnd{3.0} >>>>> \MPIupdateBegin{3.0}{278} >>>>> \constitem{MPI\_COMBINER\_HVECTOR}{\mpifunc{MPI\_TYPE\_CREATE\_HVECTOR}} >>>>> \MPIupdateEnd{3.0} >>>>> ...etc. >>>>> ----- >>>>> >>>>> If we remove all the \MPIdelete* and \MPIupdate* macros, it builds ok. If we put even one of them back, we get "Float(s) lost" and pdflatex aborts. But other than that, we're totally stumped (we've wasted 2+ hours trying to solve this so far :-( ). >>>>> >>>>> What should we do? >>>>> >>>>> -- >>>>> Jeff Squyres >>>>> jsquyres(a)cisco.com >>>>> For corporate legal information go to: >>>>> http://www.cisco.com/web/about/doing_business/legal/cri/ >>>>> >>>>> >>>>> _______________________________________________ >>>>> mpi-forum mailing list >>>>> mpi-forum(a)lists.mpi-forum.org >>>>> http://lists.mpi-forum.org/mailman/listinfo.cgi/mpi-forum >>>> >>>> William Gropp >>>> Director, Parallel Computing Institute >>>> Deputy Director for Research >>>> Institute for Advanced Computing Applications and Technologies >>>> Paul and Cynthia Saylor Professor of Computer Science >>>> University of Illinois Urbana-Champaign >>>> >>>> >>>> >>>> _______________________________________________ >>>> mpi-forum mailing list >>>> mpi-forum(a)lists.mpi-forum.org >>>> http://lists.mpi-forum.org/mailman/listinfo.cgi/mpi-forum >>> >>> William Gropp >>> Director, Parallel Computing Institute >>> Deputy Director for Research >>> Institute for Advanced Computing Applications and Technologies >>> Paul and Cynthia Saylor Professor of Computer Science >>> University of Illinois Urbana-Champaign >>> >>> >>> >>> _______________________________________________ >>> mpi-forum mailing list >>> mpi-forum(a)lists.mpi-forum.org >>> http://lists.mpi-forum.org/mailman/listinfo.cgi/mpi-forum >> >> >> -- >> Jeff Squyres >> jsquyres(a)cisco.com >> For corporate legal information go to: >> http://www.cisco.com/web/about/doing_business/legal/cri/ >> >> >> _______________________________________________ >> mpi-forum mailing list >> mpi-forum(a)lists.mpi-forum.org >> http://lists.mpi-forum.org/mailman/listinfo.cgi/mpi-forum > > William Gropp > Director, Parallel Computing Institute > Deputy Director for Research > Institute for Advanced Computing Applications and Technologies > Paul and Cynthia Saylor Professor of Computer Science > University of Illinois Urbana-Champaign > > > > _______________________________________________ > mpi-forum mailing list > mpi-forum(a)lists.mpi-forum.org > http://lists.mpi-forum.org/mailman/listinfo.cgi/mpi-forum -- Jeff Squyres jsquyres(a)cisco.com For corporate legal information go to: http://www.cisco.com/web/about/doing_business/legal/cri/

1 0

Re: [Mpi-forum] (Continued) Latex woes
by William Gropp 10 Jun '11

10 Jun '11

I've committed it in one place in the approved branch, but can't fix it in the bw-compat ticket 278 branch. Bill On Jun 10, 2011, at 7:08 AM, Jeff Squyres wrote: > Thanks! > > Is there an SVN permissions problem such that you can't commit? > (assumedly to the /minutiae branch, since the MPIreplace* (etc.) > macros are not on the /approved branch...?) > > > On Jun 9, 2011, at 8:23 PM, William Gropp wrote: > >> Attached is the diff of the fix. I can't commit it, so one of you >> will need to. >> >> Bill >> >> <fix> >> On Jun 9, 2011, at 12:20 PM, William Gropp wrote: >> >>> Floats lost is often due to a limitation in LaTeX. One possible >>> issue is that the the change markers use floats, and so do >>> tables. LaTeX isn't very happy with floats within floats. I'm in >>> a meeting today and can't look at it until later, but one solution >>> may be to inline the ticket info in these cases. >>> >>> Bill >>> >>> On Jun 9, 2011, at 11:30 AM, Jeff Squyres wrote: >>> >>>> Bill -- >>>> >>>> We're still having latex woes in ticket 278. Can you help? >>>> >>>> We replaced all the \MPIreplace macros with deletebegin/end and >>>> updatebegin/end, but we're still getting "Float(s) lost". It >>>> seems to have to do with a table in datatype.tex. The table >>>> looks like this: >>>> >>>> ----- >>>> \begin{table}[ht] >>>> \begin{center} >>>> \begin{constlist} >>>> \constitem{MPI\_COMBINER\_NAMED}{a named predefined datatype} >>>> ...etc. >>>> \MPIdeleteBegin{3.0}{278} >>>> \constitem{MPI\_COMBINER\_HVECTOR\_INTEGER}{\mpifunc{MPI\_TYPE >>>> \_HVECTOR} from Fortran} >>>> \constitem{MPI\_COMBINER\_HVECTOR}{\mpifunc{MPI\_TYPE\_HVECTOR} >>>> from C or C++} >>>> \constskipitem{}{\ \ and in some case Fortran} >>>> \constskipitem{}{\ \ or \mpifunc{MPI\_TYPE\_CREATE\_HVECTOR}} >>>> \MPIdeleteEnd{3.0} >>>> \MPIupdateBegin{3.0}{278} >>>> \constitem{MPI\_COMBINER\_HVECTOR}{\mpifunc{MPI\_TYPE\_CREATE >>>> \_HVECTOR}} >>>> \MPIupdateEnd{3.0} >>>> ...etc. >>>> ----- >>>> >>>> If we remove all the \MPIdelete* and \MPIupdate* macros, it >>>> builds ok. If we put even one of them back, we get "Float(s) >>>> lost" and pdflatex aborts. But other than that, we're totally >>>> stumped (we've wasted 2+ hours trying to solve this so far :-( ). >>>> >>>> What should we do? >>>> >>>> -- >>>> Jeff Squyres >>>> jsquyres(a)cisco.com >>>> For corporate legal information go to: >>>> http://www.cisco.com/web/about/doing_business/legal/cri/ >>>> >>>> >>>> _______________________________________________ >>>> mpi-forum mailing list >>>> mpi-forum(a)lists.mpi-forum.org >>>> http://lists.mpi-forum.org/mailman/listinfo.cgi/mpi-forum >>> >>> William Gropp >>> Director, Parallel Computing Institute >>> Deputy Director for Research >>> Institute for Advanced Computing Applications and Technologies >>> Paul and Cynthia Saylor Professor of Computer Science >>> University of Illinois Urbana-Champaign >>> >>> >>> >>> _______________________________________________ >>> mpi-forum mailing list >>> mpi-forum(a)lists.mpi-forum.org >>> http://lists.mpi-forum.org/mailman/listinfo.cgi/mpi-forum >> >> William Gropp >> Director, Parallel Computing Institute >> Deputy Director for Research >> Institute for Advanced Computing Applications and Technologies >> Paul and Cynthia Saylor Professor of Computer Science >> University of Illinois Urbana-Champaign >> >> >> >> _______________________________________________ >> mpi-forum mailing list >> mpi-forum(a)lists.mpi-forum.org >> http://lists.mpi-forum.org/mailman/listinfo.cgi/mpi-forum > > > -- > Jeff Squyres > jsquyres(a)cisco.com > For corporate legal information go to: > http://www.cisco.com/web/about/doing_business/legal/cri/ > > > _______________________________________________ > mpi-forum mailing list > mpi-forum(a)lists.mpi-forum.org > http://lists.mpi-forum.org/mailman/listinfo.cgi/mpi-forum William Gropp Director, Parallel Computing Institute Deputy Director for Research Institute for Advanced Computing Applications and Technologies Paul and Cynthia Saylor Professor of Computer Science University of Illinois Urbana-Champaign

1 0