[OMPI devel] RFC: Remove old MCA parameter system from trunk
What: The mca_base_var_* API replaces the mca_base_var_* API. A support layer for the old API currently resides in the trunk. I will remove the layer from the trunk and replace any remaining mca_base_param_* calls with their equivalents in the new API. The support layer will remain for the 1.7/1.8 release series. When: This RFC is a heads up. I will remove the old API on Monday, May 20, 2013. -Nathan Hjelm HPC-3, LANL
[OMPI devel] 1.7.2rc1
Hi folks We have posted the first release candidate for the 1.7.2 release in the usual place: http://www.open-mpi.org/software/ompi/v1.7/ Please put it thru the wringer to help us validate it prior to release. Thanks Ralph
[OMPI devel] RFC: dead code removal
WHAT: Remove a bunch of "set but not used" variables / dead code WHY: Because it's dead code WHERE: All over, but NOT the BTL ALLOC macros (per prior argu^H^H^H^Hdiscussion) WHEN: Tomorrow (16 May 2013), COB More detail: gcc 4.7.x squawks a lot about "set but unused" variables. I took a sweep through and removed a bunch of them -- they're all obviously dead code. I did *not*, however, remove the setting of rc in the various BTL/OOB ALLOC_FRAG macros, per prior disagreements in emails about this. Perhaps someone else will find a compromise for that someday -- this patch is not about fixing those warnings. This patch is only about removing the obvious set-but-really-never-used variables. Short timeout because this is actually pretty trivial, but it does touch other people's code, so I wanted people to see it / get a heads-up before I committed. Patch attached. -- Jeff Squyres jsquy...@cisco.com For corporate legal information go to: http://www.cisco.com/web/about/doing_business/legal/cri/ dead-code-removal-trunk.diff Description: dead-code-removal-trunk.diff
Re: [OMPI devel] RFC: dead code removal
Hmmm...some of this doesn't look right to me. It could be that some of the code changed and stale things didn't get removed, but the snippets of logic in your patch raise alarms in some cases. Can you allow a bit more time? I need to apply the patch and actually look at the total code path to understand *why* some of these variables are no longer being used. My fear is that there are cmd line options that may not be working correctly (but rarely get used/tested) because (a) the variable is correct, but (b) somehow the rest of the code is in error. On May 15, 2013, at 5:24 PM, Jeff Squyres (jsquyres) wrote: > WHAT: Remove a bunch of "set but not used" variables / dead code > > WHY: Because it's dead code > > WHERE: All over, but NOT the BTL ALLOC macros (per prior > argu^H^H^H^Hdiscussion) > > WHEN: Tomorrow (16 May 2013), COB > > More detail: > > gcc 4.7.x squawks a lot about "set but unused" variables. I took a sweep > through and removed a bunch of them -- they're all obviously dead code. > > I did *not*, however, remove the setting of rc in the various BTL/OOB > ALLOC_FRAG macros, per prior disagreements in emails about this. Perhaps > someone else will find a compromise for that someday -- this patch is not > about fixing those warnings. This patch is only about removing the obvious > set-but-really-never-used variables. > > Short timeout because this is actually pretty trivial, but it does touch > other people's code, so I wanted people to see it / get a heads-up before I > committed. Patch attached. > > -- > Jeff Squyres > jsquy...@cisco.com > For corporate legal information go to: > http://www.cisco.com/web/about/doing_business/legal/cri/ > ___ > devel mailing list > de...@open-mpi.org > http://www.open-mpi.org/mailman/listinfo.cgi/devel
Re: [OMPI devel] RFC: dead code removal
Sure, no problem. On May 15, 2013, at 8:41 PM, Ralph Castain wrote: > Hmmm...some of this doesn't look right to me. It could be that some of the > code changed and stale things didn't get removed, but the snippets of logic > in your patch raise alarms in some cases. > > Can you allow a bit more time? I need to apply the patch and actually look at > the total code path to understand *why* some of these variables are no longer > being used. My fear is that there are cmd line options that may not be > working correctly (but rarely get used/tested) because (a) the variable is > correct, but (b) somehow the rest of the code is in error. > > > On May 15, 2013, at 5:24 PM, Jeff Squyres (jsquyres) > wrote: > >> WHAT: Remove a bunch of "set but not used" variables / dead code >> >> WHY: Because it's dead code >> >> WHERE: All over, but NOT the BTL ALLOC macros (per prior >> argu^H^H^H^Hdiscussion) >> >> WHEN: Tomorrow (16 May 2013), COB >> >> More detail: >> >> gcc 4.7.x squawks a lot about "set but unused" variables. I took a sweep >> through and removed a bunch of them -- they're all obviously dead code. >> >> I did *not*, however, remove the setting of rc in the various BTL/OOB >> ALLOC_FRAG macros, per prior disagreements in emails about this. Perhaps >> someone else will find a compromise for that someday -- this patch is not >> about fixing those warnings. This patch is only about removing the obvious >> set-but-really-never-used variables. >> >> Short timeout because this is actually pretty trivial, but it does touch >> other people's code, so I wanted people to see it / get a heads-up before I >> committed. Patch attached. >> >> -- >> Jeff Squyres >> jsquy...@cisco.com >> For corporate legal information go to: >> http://www.cisco.com/web/about/doing_business/legal/cri/ >> ___ >> devel mailing list >> de...@open-mpi.org >> http://www.open-mpi.org/mailman/listinfo.cgi/devel > > > ___ > devel mailing list > de...@open-mpi.org > http://www.open-mpi.org/mailman/listinfo.cgi/devel -- Jeff Squyres jsquy...@cisco.com For corporate legal information go to: http://www.cisco.com/web/about/doing_business/legal/cri/
