April 2012 - petsc-dev - mailman3.cels.anl.gov

Re: [petsc-dev] Detecting MPI_C_DOUBLE_COMPLEX when clanguage!='C'
by Jed Brown 24 Apr '12

24 Apr '12

The alternative is that 1. Doing nothing means it gets lost without resolution. 2. The only way to be notified of action is to subscribe to the mailing list (and not all fixes are mentioned there) which is high volume if all you care about is a particular bug. > On Apr 24, 2012 8:05 AM, "Barry Smith" <bsmith(a)mcs.anl.gov> wrote: >> >> >> From that page "status changed from new to closed" >> >> This is why I absolutely hate the idea of using "ticket" systems for open source projects. The goal becomes "close the tickets at all cost using any lame ass excuse you can" instead of actually thinking about the problem and coming up with a satisfactory solution. >> >> BTW: How is OpenMPI on complex for Windows and C++ and stuff? >> >> Barry >> >> On Apr 24, 2012, at 7:24 AM, Matthew Knepley wrote: >> >> > On Tue, Apr 24, 2012 at 8:19 AM, Jed Brown <jedbrown(a)mcs.anl.gov> wrote: >> > On Tue, Apr 24, 2012 at 07:15, Matthew Knepley <knepley(a)gmail.com> wrote: >> > Of course don't fix it. Who do these guys work for, Sandia? >> > >> > They could fix it as an extension in MPICH2, but it would have to be added to the standard for us to rely on it. Considering that we still can't use anything past MPI-1.1, we won't be able to remove the hacks for a long time. This is not to say that it isn't worth fixing, just that your daughter will be in college by the time the fix can be assumed by callers. >> > >> > MPICH used to lead by just doing good, useful things. The embarrassed standards committee >> > would then adopt them. Now this is reversed? >> > >> > Matt >> > >> > -- >> > What most experimenters take for granted before they begin their experiments is infinitely more interesting than any results to which their experiments lead. >> > -- Norbert Wiener >>

1 0

Re: [petsc-dev] Detecting MPI_C_DOUBLE_COMPLEX when clanguage!='C'
by Barry Smith 24 Apr '12

24 Apr '12

From that page "status changed from new to closed" This is why I absolutely hate the idea of using "ticket" systems for open source projects. The goal becomes "close the tickets at all cost using any lame ass excuse you can" instead of actually thinking about the problem and coming up with a satisfactory solution. BTW: How is OpenMPI on complex for Windows and C++ and stuff? Barry On Apr 24, 2012, at 7:24 AM, Matthew Knepley wrote: > On Tue, Apr 24, 2012 at 8:19 AM, Jed Brown <jedbrown(a)mcs.anl.gov> wrote: > On Tue, Apr 24, 2012 at 07:15, Matthew Knepley <knepley(a)gmail.com> wrote: > Of course don't fix it. Who do these guys work for, Sandia? > > They could fix it as an extension in MPICH2, but it would have to be added to the standard for us to rely on it. Considering that we still can't use anything past MPI-1.1, we won't be able to remove the hacks for a long time. This is not to say that it isn't worth fixing, just that your daughter will be in college by the time the fix can be assumed by callers. > > MPICH used to lead by just doing good, useful things. The embarrassed standards committee > would then adopt them. Now this is reversed? > > Matt > > -- > What most experimenters take for granted before they begin their experiments is infinitely more interesting than any results to which their experiments lead. > -- Norbert Wiener

1 0

Re: [petsc-dev] Detecting MPI_C_DOUBLE_COMPLEX when clanguage!='C'
by Matthew Knepley 24 Apr '12

24 Apr '12

On Tue, Apr 24, 2012 at 8:19 AM, Jed Brown <jedbrown(a)mcs.anl.gov> wrote: > On Tue, Apr 24, 2012 at 07:15, Matthew Knepley <knepley(a)gmail.com> wrote: > >> Of course don't fix it. Who do these guys work for, Sandia? > > > They could fix it as an extension in MPICH2, but it would have to be added > to the standard for us to rely on it. Considering that we still can't use > anything past MPI-1.1, we won't be able to remove the hacks for a long > time. This is not to say that it isn't worth fixing, just that your > daughter will be in college by the time the fix can be assumed by callers. > MPICH used to lead by just doing good, useful things. The embarrassed standards committee would then adopt them. Now this is reversed? Matt -- What most experimenters take for granted before they begin their experiments is infinitely more interesting than any results to which their experiments lead. -- Norbert Wiener

1 0

Re: [petsc-dev] Detecting MPI_C_DOUBLE_COMPLEX when clanguage!='C'
by Jed Brown 24 Apr '12

24 Apr '12

On Tue, Apr 24, 2012 at 07:15, Matthew Knepley <knepley(a)gmail.com> wrote: > Of course don't fix it. Who do these guys work for, Sandia? They could fix it as an extension in MPICH2, but it would have to be added to the standard for us to rely on it. Considering that we still can't use anything past MPI-1.1, we won't be able to remove the hacks for a long time. This is not to say that it isn't worth fixing, just that your daughter will be in college by the time the fix can be assumed by callers.

1 0

Re: [petsc-dev] Detecting MPI_C_DOUBLE_COMPLEX when clanguage!='C'
by Matthew Knepley 24 Apr '12

24 Apr '12

On Tue, Apr 24, 2012 at 7:42 AM, Jed Brown <jedbrown(a)mcs.anl.gov> wrote: > On Tue, Apr 24, 2012 at 05:23, Matthew Knepley <knepley(a)gmail.com> wrote: > >> On Mon, Apr 23, 2012 at 10:05 PM, Jed Brown <jedbrown(a)mcs.anl.gov> wrote: >> >>> On Mon, Apr 23, 2012 at 20:59, Satish Balay <balay(a)mcs.anl.gov> wrote: >>> >>>> Well it didn't work on mpich/windows - and as per the following >>>> discussion - I made >>>> that change. >>>> >>>> http://trac.mcs.anl.gov/projects/mpich2/ticket/1525 >>>> >>> >>> How horrible. >>> >> >> Classic pass-the--buck response. Still want to tell me they deserve their >> money? >> > > As Rajeev says, this looks like a mistake that was made 15 years ago. > Of course don't fix it. Who do these guys work for, Sandia? Matt -- What most experimenters take for granted before they begin their experiments is infinitely more interesting than any results to which their experiments lead. -- Norbert Wiener

1 0

Re: [petsc-dev] Detecting MPI_C_DOUBLE_COMPLEX when clanguage!='C'
by Jed Brown 24 Apr '12

24 Apr '12

On Tue, Apr 24, 2012 at 05:23, Matthew Knepley <knepley(a)gmail.com> wrote: > On Mon, Apr 23, 2012 at 10:05 PM, Jed Brown <jedbrown(a)mcs.anl.gov> wrote: > >> On Mon, Apr 23, 2012 at 20:59, Satish Balay <balay(a)mcs.anl.gov> wrote: >> >>> Well it didn't work on mpich/windows - and as per the following >>> discussion - I made >>> that change. >>> >>> http://trac.mcs.anl.gov/projects/mpich2/ticket/1525 >>> >> >> How horrible. >> > > Classic pass-the--buck response. Still want to tell me they deserve their > money? > As Rajeev says, this looks like a mistake that was made 15 years ago.

1 0

Re: [petsc-dev] Detecting MPI_C_DOUBLE_COMPLEX when clanguage!='C'
by Matthew Knepley 24 Apr '12

24 Apr '12

On Mon, Apr 23, 2012 at 10:05 PM, Jed Brown <jedbrown(a)mcs.anl.gov> wrote: > On Mon, Apr 23, 2012 at 20:59, Satish Balay <balay(a)mcs.anl.gov> wrote: > >> Well it didn't work on mpich/windows - and as per the following >> discussion - I made >> that change. >> >> http://trac.mcs.anl.gov/projects/mpich2/ticket/1525 >> > > How horrible. > Classic pass-the--buck response. Still want to tell me they deserve their money? Matt -- What most experimenters take for granted before they begin their experiments is infinitely more interesting than any results to which their experiments lead. -- Norbert Wiener

1 0

Re: [petsc-dev] Detecting MPI_C_DOUBLE_COMPLEX when clanguage!='C'
by Jed Brown 23 Apr '12

23 Apr '12

On Mon, Apr 23, 2012 at 20:59, Satish Balay <balay(a)mcs.anl.gov> wrote: > Well it didn't work on mpich/windows - and as per the following discussion > - I made > that change. > > http://trac.mcs.anl.gov/projects/mpich2/ticket/1525 > How horrible.

1 0

Re: [petsc-dev] Detecting MPI_C_DOUBLE_COMPLEX when clanguage!='C'
by Satish Balay 23 Apr '12

23 Apr '12

On Mon, 23 Apr 2012, Barry Smith wrote: > > On Apr 23, 2012, at 7:51 PM, Jed Brown wrote: > > > if self.getDefaultLanguage() == 'C': mpitypes.extend([('MPI_C_DOUBLE_COMPLEX', 'c-double-complex')]) > > > > Why don't we check for this regardless of C language? The C and C++ types are binary compatible and there is an advantage to using the predefined data types: it lets us also use the predefined MPI_Ops, which would let us use MPIU_SUM with PetscSF (and possibly use more optimized reduction implementations in other cases). > > > Though this may always work are they technically really binary compatible and does MPI (the standard) really support complex in C++ since it is not the same as complex in C? > > Yes this mostly academic bullshit and we should try to see if it works with C++ in BuildSystem also. Well it didn't work on mpich/windows - and as per the following discussion - I made that change. http://trac.mcs.anl.gov/projects/mpich2/ticket/1525 Satish

1 0

Re: [petsc-dev] Detecting MPI_C_DOUBLE_COMPLEX when clanguage!='C'
by Barry Smith 23 Apr '12

23 Apr '12

On Apr 23, 2012, at 7:51 PM, Jed Brown wrote: > if self.getDefaultLanguage() == 'C': mpitypes.extend([('MPI_C_DOUBLE_COMPLEX', 'c-double-complex')]) > > Why don't we check for this regardless of C language? The C and C++ types are binary compatible and there is an advantage to using the predefined data types: it lets us also use the predefined MPI_Ops, which would let us use MPIU_SUM with PetscSF (and possibly use more optimized reduction implementations in other cases). Though this may always work are they technically really binary compatible and does MPI (the standard) really support complex in C++ since it is not the same as complex in C? Yes this mostly academic bullshit and we should try to see if it works with C++ in BuildSystem also. Barry

1 0