-
Notifications
You must be signed in to change notification settings - Fork 3
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
Small clarifications in chapter 10 #13
Comments
Originally by jsquyres on 2008-09-09 16:51:55 -0500 Slightly modified some of the wiki markup in the description. |
Originally by dries on 2008-10-14 07:43:08 -0500 transferred straw vote results from wiki page to ticket |
Originally by gropp on 2008-10-26 09:56:50 -0500 Specific text is needed for this ticket. The Forum appears to want to preserve the historical information while updating the text. |
Originally by RolfRabenseifner on 2009-01-17 04:59:42 -0600 Please keep the Change-Log entry empty, i.e., the text should be only You may move your comment into the "History" section. |
Originally by gropp on 2009-02-06 08:26:53 -0600 Here's the current text: 10.2.1 Starting ProcessesMPI applications may start new processes through an interface to an external process manager, which can range from a parallel operating system (CMOST) to layered software (POE) to an rsh command (p4). In this place, I don't believe the existing examples are helpful - so I propose this alternative text: MPI applications may start new processes through an interface to an external process Comments? |
Originally by traff on 2009-02-10 18:14:10 -0600 I think the proposal by Bill is good, and should be put forward. Any takers? (are you still active, Dries?) Jesper |
Originally by gropp on 2009-02-13 16:00:44 -0600 I've updated this to use my proposed solution, and I've corrected the change log entry. This is now ready for review. |
Originally by traff on 2009-03-04 03:07:57 -0600 The phrase "a difficult situation" is perhaps not the most elegant or telling (proposed for p. 307). What about "a vicious circle" or "a contradicting situation", or even "a catch-22 situation." Otherwise I agree that references to specific (older) interfaces should be deleted (p. 294) Jesper |
Originally by RolfRabenseifner on 2009-03-24 11:06:39 -0500 My review:
My opinion (not review): I would prefer "a contradicting situation" |
Originally by jsquyres on 2009-04-02 14:34:06 -0500 Dries / Bill -- do you want this ticket to move forward in Chicago next week? |
Originally by dries on 2009-04-03 14:36:26 -0500 I agree with Bill's suggestion: replace "MPI applications may start new processes through an interface to an external process manager, which can range from a parallel operating system (CMOST) to layered software (POE) to an rsh command (p4)." with "MPI applications may start new processes through an interface to an external process manager." Obvisouly, the section header: "10.2.1 Starting Processes" can stay. I copied it into the ticket to provide some context. About "catch-22": I prefer "a vicous circle" or dropping "- catch 22" alltogether. |
Originally by jsquyres on 2009-04-04 10:48:17 -0500 Dries -- will you be updating the proposal? The modus operandi for MPI-2.2 tickets is to edit the proposal itself (you can edit the original text by scrolling down to the bottom of the ticket page and editing directly) and then we review it again. |
Originally by gropp on 2009-04-05 16:09:49 -0500 I've updated the proposal, choosing the "drop the Catch-22" as the option for the second change (and including the entire sentence for context). |
Originally by jsquyres on 2009-04-06 05:51:14 -0500 Reviewed ok. |
Originally by RolfRabenseifner on 2009-04-07 10:08:06 -0500 Okay. |
Originally by rsthakur on 2009-04-07 10:50:50 -0500 OK |
Originally by RolfRabenseifner on 2009-04-07 11:30:15 -0500 Taking Bills, Jeffs, Rajeevs, and my review, this ticket can be set to "Reviewed". |
Originally by dgsolt on 2009-07-12 01:20:38 -0500 Attachment added: |
Originally by dgsolt on 2009-07-12 01:21:11 -0500 Attachment added: |
Originally by RolfRabenseifner on 2009-08-28 17:07:11 -0500 PDF review: okay. |
Originally by bronis on 2009-08-29 18:37:20 -0500 PDF Review OK. |
Originally by traff on 2009-08-30 06:29:10 -0500 PDF reviewed, OK |
Originally by jsquyres on 2010-09-18 04:58:32 -0500 This ticket is (long-since) complete; marking it resolved/text committed. |
Originally by dries on 2008-09-08 15:22:37 -0500
Small clarifications in chapter 10Author: Dries Kimpe
Status
Sep 8 2008 Straw Vote
|| No Change || Add something || Delete || Abstain ||
|| 0 || 22 || 8 || 2 ||
Description
Chapter 10 references POE, P4 and CMOST without further explanation or
citations.
I propose to -- at the least -- add references for these abbreviations.
However, considering that POE,P4 and CMOST are outdated,
it might be better to rewrite the section and/or replace by more modern
examples.
This proposal provides references for POE, P4 and CMOST.
Also, I propose to replace 'catch-22' by something more explicit.
History
Proposed Solution
The current text reads:
MPI applications may start new processes through an interface to an external process manager, which can range from a parallel operating system (CMOST) to layered software (POE) to an rsh command (p4).
Replace this with this text:
MPI applications may start new processes through an interface to an external process manager.
The current text reads
The difficulty, of course, is that there
is no existing communication channel between them, yet they must somehow agree on a
rendezvous point where they will establish communication — Catch-22.
Replace this with this text:
The difficulty, of course, is that there
is no existing communication channel between them, yet they must somehow agree on a
rendezvous point where they will establish communication.
Impact on Implementations
Impact on Applications / Users
Alternative Solutions
Add these references:
Comment by Dick Treumann - POE is not outdated. "poe" is the process which manages the tasks of an MPI job in IBM's Parallel Environment product. It also can be the explicit job launch command. I.E. an MPI program "myprog" for Parallel Environment can be launched by either "poe myprog " or simply "myprog ". IBM PE also supports "mpiexec". In any case there will be a "poe" process running for each MPI job. (I am not sure what the phrase "layered software" is intended to convey - perhaps that POE is built on OS services rather than an is OS service itself). A PE implementation of spawn would certainly involve interaction with job's "poe".
Entry for the Change Log
NA
The text was updated successfully, but these errors were encountered: