May 2012 - mpi-forum - mailman3.cels.anl.gov

Re: [Mpi-forum] Voting results
by Douglas Miller 30 May '12

30 May '12

I have to "second" Fab's objection. this does seem wrong - even though it doesn't help my ticket to pass... _______________________________________________ Douglas Miller BlueGene Messaging Development IBM Corp., Rochester, MN USA Bldg 030-2 A401 dougmill(a)us.ibm.com Douglas Miller/Rochester/IBM Fab Tillier <ftillier@microso ft.com> To Sent by: Main MPI Forum mailing list mpi-forum-bounces <mpi-forum(a)lists.mpi-forum.org>, @lists.mpi-forum. cc org Subject Re: [Mpi-forum] Voting results 05/30/2012 01:51 PM Please respond to Main MPI Forum mailing list <mpi-forum@lists. mpi-forum.org> Jeff Squyres wrote on Wed, 30 May 2012 at 11:36:09 > 2. The definition of "simple majority" was changed from how I have > computed whether ballots passed or failed in the past. I don't know offhand > how past ballot results would have fared with the new definition; I am > guessing that their results wouldn't have changed because most past ballots > were not as close as some of the ones from this week. > > From my understanding, "simple majority" (i.e., what a vote needs to pass) > was defined as the following: > > floor(total_eligible_orgs_attending / 2) + 1 "yes" votes > Meaning: abstains and misses count as "not yes", or (effectively) "no". > > *** With these rules, I see no meaning for "abstain" (or "miss"). There is > effectively only "yes" and "no". > *** Meaning: everyone who thought they were abstaining at this past > meeting were actually voting "no". > > I understand that this was discussed in Japan and everyone in the room > agreed to these rules. ***It is not what I would have advocated***, but I > was not there. :-\ > > In all prior meetings, I used the following computation to determine if > a ballot passed: > > floor(total_yes_and_no_votes / 2) + 1 "yes" votes > or, effectively: > > more "yes" votes than "no" votes > Meaning: abstains and misses do not count towards the result. IMO this kind of change is not something that should happen in a single meeting. Just like we don't make large changes to the standard in a single meeting, I feel very strongly that the MPI Forum follow the same kind of process in making such significant rule changes as we do with tickets. To be clear, I believe that this change should have been brought up one meeting, voted in the next, and voted a second time to pass in the 3rd meeting. Yes, it would take time, but bylaw changes should not be undertaken lightly. The fact that some votes were still recorded as 'abstain' is an indication that this bylaw change was half baked. -Fab _______________________________________________ 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] Voting results
by Jeff Hammond 30 May '12

30 May '12

On Wed, May 30, 2012 at 1:51 PM, Fab Tillier <ftillier(a)microsoft.com> wrote: > Jeff Squyres wrote on Wed, 30 May 2012 at 11:36:09 > >> 2. The definition of "simple majority" was changed from how I have >> computed whether ballots passed or failed in the past. ï¿½I don't know offhand >> how past ballot results would have fared with the new definition; I am >> guessing that their results wouldn't have changed because most past ballots >> were not as close as some of the ones from this week. >> >> From my understanding, "simple majority" (i.e., what a vote needs to pass) >> was defined as the following: >> >> ï¿½ ï¿½ floor(total_eligible_orgs_attending / 2) + 1 "yes" votes >> Meaning: abstains and misses count as "not yes", or (effectively) "no". >> >> *** With these rules, I see no meaning for "abstain" (or "miss"). ï¿½There is >> effectively only "yes" and "no". >> *** Meaning: everyone who thought they were abstaining at this past >> meeting were actually voting "no". >> >> I understand that this was discussed in Japan and everyone in the room >> agreed to these rules. ï¿½***It is not what I would have advocated***, but I >> was not there. ï¿½:-\ >> >> In all prior meetings, I used the following computation to determine if >> a ballot passed: >> >> ï¿½ ï¿½ floor(total_yes_and_no_votes / 2) + 1 "yes" votes >> or, effectively: >> >> ï¿½ ï¿½ more "yes" votes than "no" votes >> Meaning: abstains and misses do not count towards the result. > > IMO this kind of change is not something that should happen in a single meeting. ï¿½Just like we don't make large changes to the standard in a single meeting, I feel very strongly that the MPI Forum follow the same kind of process in making such significant rule changes as we do with tickets. ï¿½To be clear, I believe that this change should have been brought up one meeting, voted in the next, and voted a second time to pass in the 3rd meeting. ï¿½Yes, it would take time, but bylaw changes should not be undertaken lightly. > > The fact that some votes were still recorded as 'abstain' is an indication that this bylaw change was half baked. > > -Fab Especially when the meeting is attended by so few people due to the location. It seems like a weasel tactic to pick a remote location to change the by-laws with a single vote. Jeff -- Jeff Hammond Argonne Leadership Computing Facility University of Chicago Computation Institute jhammond(a)alcf.anl.gov / (630) 252-5381 http://www.linkedin.com/in/jeffhammond https://wiki.alcf.anl.gov/parts/index.php/User:Jhammond

1 0

Re: [Mpi-forum] Voting results
by Darius Buntinas 30 May '12

30 May '12

Would any of the tickets that were voted down with the Japan rules have passed if we used the US/Europe rules? -d On May 30, 2012, at 1:51 PM, Fab Tillier wrote: > Jeff Squyres wrote on Wed, 30 May 2012 at 11:36:09 > >> 2. The definition of "simple majority" was changed from how I have >> computed whether ballots passed or failed in the past. I don't know offhand >> how past ballot results would have fared with the new definition; I am >> guessing that their results wouldn't have changed because most past ballots >> were not as close as some of the ones from this week. >> >> From my understanding, "simple majority" (i.e., what a vote needs to pass) >> was defined as the following: >> >> floor(total_eligible_orgs_attending / 2) + 1 "yes" votes >> Meaning: abstains and misses count as "not yes", or (effectively) "no". >> >> *** With these rules, I see no meaning for "abstain" (or "miss"). There is >> effectively only "yes" and "no". >> *** Meaning: everyone who thought they were abstaining at this past >> meeting were actually voting "no". >> >> I understand that this was discussed in Japan and everyone in the room >> agreed to these rules. ***It is not what I would have advocated***, but I >> was not there. :-\ >> >> In all prior meetings, I used the following computation to determine if >> a ballot passed: >> >> floor(total_yes_and_no_votes / 2) + 1 "yes" votes >> or, effectively: >> >> more "yes" votes than "no" votes >> Meaning: abstains and misses do not count towards the result. > > IMO this kind of change is not something that should happen in a single meeting. Just like we don't make large changes to the standard in a single meeting, I feel very strongly that the MPI Forum follow the same kind of process in making such significant rule changes as we do with tickets. To be clear, I believe that this change should have been brought up one meeting, voted in the next, and voted a second time to pass in the 3rd meeting. Yes, it would take time, but bylaw changes should not be undertaken lightly. > > The fact that some votes were still recorded as 'abstain' is an indication that this bylaw change was half baked. > > -Fab > > > _______________________________________________ > 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] Voting results
by Fab Tillier 30 May '12

30 May '12

Jeff Squyres wrote on Wed, 30 May 2012 at 11:36:09 > 2. The definition of "simple majority" was changed from how I have > computed whether ballots passed or failed in the past. I don't know offhand > how past ballot results would have fared with the new definition; I am > guessing that their results wouldn't have changed because most past ballots > were not as close as some of the ones from this week. > > From my understanding, "simple majority" (i.e., what a vote needs to pass) > was defined as the following: > > floor(total_eligible_orgs_attending / 2) + 1 "yes" votes > Meaning: abstains and misses count as "not yes", or (effectively) "no". > > *** With these rules, I see no meaning for "abstain" (or "miss"). There is > effectively only "yes" and "no". > *** Meaning: everyone who thought they were abstaining at this past > meeting were actually voting "no". > > I understand that this was discussed in Japan and everyone in the room > agreed to these rules. ***It is not what I would have advocated***, but I > was not there. :-\ > > In all prior meetings, I used the following computation to determine if > a ballot passed: > > floor(total_yes_and_no_votes / 2) + 1 "yes" votes > or, effectively: > > more "yes" votes than "no" votes > Meaning: abstains and misses do not count towards the result. IMO this kind of change is not something that should happen in a single meeting. Just like we don't make large changes to the standard in a single meeting, I feel very strongly that the MPI Forum follow the same kind of process in making such significant rule changes as we do with tickets. To be clear, I believe that this change should have been brought up one meeting, voted in the next, and voted a second time to pass in the 3rd meeting. Yes, it would take time, but bylaw changes should not be undertaken lightly. The fact that some votes were still recorded as 'abstain' is an indication that this bylaw change was half baked. -Fab

1 0

MPI Forum services migration: part 1
by Jeff Squyres 30 May '12

30 May '12

Be advised that I have received the following message from our Indiana U. hosting provider. SVN/Trac will be unavailable during the time frame described below. More migrations will follow (e.g., web services) in the coming weeks; stay tuned. ----- We are planning to move all the services of svn.open-mpi.org and svn.mpi-forum.org to the new server (tiger.crest.iu.edu) on June 1st, 2012. Date: June 1st, 2012 Time: - 6:00am-10:00am Pacific US time - 7:00am-11:00am Mountain US time - 8:00am-12:00pm Central US time - 9:00am-1:00pm Eastern US time - 2:00pm-6:00pm GMT * The services to migrate from sourcehaven to tiger - subversion - trac - svnmanager - opengrok During the migration, the above services will not be available. ----- -- Jeff Squyres jsquyres(a)cisco.com For corporate legal information go to: http://www.cisco.com/web/about/doing_business/legal/cri/ -- Jeff Squyres jsquyres(a)cisco.com For corporate legal information go to: http://www.cisco.com/web/about/doing_business/legal/cri/

1 0

Voting results
by Jeff Squyres 30 May '12

30 May '12

Voting results from the May 2012 Japan MPI Forum meeting are now up: http://meetings.mpi-forum.org/secretary/2012/05/votes.php Interesting points: 1. 3 tickets were withdrawn: - https://svn.mpi-forum.org/trac/mpi-forum-web/ticket/194 Ensure MPI_Dims_create is really suitable for MPI_Cart_create - https://svn.mpi-forum.org/trac/mpi-forum-web/ticket/326 MPI3 Fault Tolerance - Files - https://svn.mpi-forum.org/trac/mpi-forum-web/ticket/327 MPI3 Fault Tolerance - Dynamic process management 2. The definition of "simple majority" was changed from how I have computed whether ballots passed or failed in the past. I don't know offhand how past ballot results would have fared with the new definition; I am guessing that their results wouldn't have changed because most past ballots were not as close as some of the ones from this week. >From my understanding, "simple majority" (i.e., what a vote needs to pass) was defined as the following: floor(total_eligible_orgs_attending / 2) + 1 "yes" votes Meaning: abstains and misses count as "not yes", or (effectively) "no". *** With these rules, I see no meaning for "abstain" (or "miss"). There is effectively only "yes" and "no". *** Meaning: everyone who thought they were abstaining at this past meeting were actually voting "no". I understand that this was discussed in Japan and everyone in the room agreed to these rules. ***It is not what I would have advocated***, but I was not there. :-\ In all prior meetings, I used the following computation to determine if a ballot passed: floor(total_yes_and_no_votes / 2) + 1 "yes" votes or, effectively: more "yes" votes than "no" votes Meaning: abstains and misses do not count towards the result. 3. 16 organizations were eligible to vote at this meeting. 4. In this meeting, 12 first votes passed. 8 of those 12 had good consensus (max of 3 (no+abstain)), but 4 did not: - #278 (fix deprecated examples): 6 abstains - #281 (remove C++): 5 no votes - #303 (removed interface chapter): 5 no votes, 2 abstains - #313 (INIT/FINALIZE): 1 no vote, 4 abstains 5. In this meeting, 5 first votes failed: - https://svn.mpi-forum.org/trac/mpi-forum-web/ticket/195 Topology awareness in MPI_Dims_create - https://svn.mpi-forum.org/trac/mpi-forum-web/ticket/217 MPI3 Hybrid Programming: Proposal for Helper Threads - https://svn.mpi-forum.org/trac/mpi-forum-web/ticket/273 Add Immediate versions of nonblocking collective I/O routines - https://svn.mpi-forum.org/trac/mpi-forum-web/ticket/310 clarify MPI behavior when multiple MPI processes run in the same address space - https://svn.mpi-forum.org/trac/mpi-forum-web/ticket/323 User-Level Failure Mitigation 6. In this meeting, 8 second votes passed (i.e., all of them). Good consensus on all of them (max of 3 (no+abstain) votes). -- Jeff Squyres jsquyres(a)cisco.com For corporate legal information go to: http://www.cisco.com/web/about/doing_business/legal/cri/

1 0

Slides / Formal readings
by Jeff Squyres 30 May '12

30 May '12

1. Please send me your slides from the Japan meeting. (I have slides from Torsten already) 2. Were there any formal readings in the Japan meeting? I ask because I'm filling in http://meetings.mpi-forum.org/secretary/2012/05/votes.php. Thanks. -- 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] Migration of Forum SVN and Trac services
by Jeff Squyres 30 May '12

30 May '12

Sweet -- thank for the confirmation! On May 30, 2012, at 7:37 AM, Torsten Hoefler wrote: > Jeff, > > Rich asked at the Forum and nobody had any objections to Friday. > > All the Best, > Torsten > > On Wed, May 30, 2012 at 07:34:51AM -0400, Jeff Squyres wrote: >> I heard back from only 1 person regarding this (who said Friday was ok), so I'm going to assume that Friday, June 1, 2012 is an acceptable planned outage for the MPI Forum SVN and Trac services. >> >> >> On May 29, 2012, at 9:14 AM, Jeff Squyres wrote: >> >>> Heads up: Indiana University would like to migrate the Forum's Subversion and Trac services to new servers sometime in the immediate future. The purpose of the migration is to retire some (really) old server hardware. Delivered SVN and Trac performance probably won't be significantly impacted. >>> >>> Doing the migration while you are all at the Japan meeting -- and actively using both SVN and Trac -- is obviously a bad idea. >>> >>> *** Would it be acceptable to move everything on Friday morning, US Eastern time (June 1, 2012)? I *assume* that you'll all be flying back at that point, or asleep, or otherwise signed out for the weekend. >>> >>> -- >>> 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 >> >> >> -- >> 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 >> > > -- > ### qreharg rug ebs fv crryF ------------- http://www.unixer.de/ ----- > Torsten Hoefler | Performance Modeling and Simulation Lead > Blue Waters Directorate | University of Illinois (UIUC) > 1205 W Clark Street | Urbana, IL, 61801 > NCSA Building | +01 (217) 244-7736 > _______________________________________________ > 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] Migration of Forum SVN and Trac services
by Torsten Hoefler 30 May '12

30 May '12

Jeff, Rich asked at the Forum and nobody had any objections to Friday. All the Best, Torsten On Wed, May 30, 2012 at 07:34:51AM -0400, Jeff Squyres wrote: > I heard back from only 1 person regarding this (who said Friday was ok), so I'm going to assume that Friday, June 1, 2012 is an acceptable planned outage for the MPI Forum SVN and Trac services. > > > On May 29, 2012, at 9:14 AM, Jeff Squyres wrote: > > > Heads up: Indiana University would like to migrate the Forum's Subversion and Trac services to new servers sometime in the immediate future. The purpose of the migration is to retire some (really) old server hardware. Delivered SVN and Trac performance probably won't be significantly impacted. > > > > Doing the migration while you are all at the Japan meeting -- and actively using both SVN and Trac -- is obviously a bad idea. > > > > *** Would it be acceptable to move everything on Friday morning, US Eastern time (June 1, 2012)? I *assume* that you'll all be flying back at that point, or asleep, or otherwise signed out for the weekend. > > > > -- > > 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 > > > -- > 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 > -- ### qreharg rug ebs fv crryF ------------- http://www.unixer.de/ ----- Torsten Hoefler | Performance Modeling and Simulation Lead Blue Waters Directorate | University of Illinois (UIUC) 1205 W Clark Street | Urbana, IL, 61801 NCSA Building | +01 (217) 244-7736

1 0

Re: [Mpi-forum] Migration of Forum SVN and Trac services
by Jeff Squyres 30 May '12

30 May '12

I heard back from only 1 person regarding this (who said Friday was ok), so I'm going to assume that Friday, June 1, 2012 is an acceptable planned outage for the MPI Forum SVN and Trac services. On May 29, 2012, at 9:14 AM, Jeff Squyres wrote: > Heads up: Indiana University would like to migrate the Forum's Subversion and Trac services to new servers sometime in the immediate future. The purpose of the migration is to retire some (really) old server hardware. Delivered SVN and Trac performance probably won't be significantly impacted. > > Doing the migration while you are all at the Japan meeting -- and actively using both SVN and Trac -- is obviously a bad idea. > > *** Would it be acceptable to move everything on Friday morning, US Eastern time (June 1, 2012)? I *assume* that you'll all be flying back at that point, or asleep, or otherwise signed out for the weekend. > > -- > 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 -- Jeff Squyres jsquyres(a)cisco.com For corporate legal information go to: http://www.cisco.com/web/about/doing_business/legal/cri/

1 0