March 2011 - petsc-dev - mailman3.cels.anl.gov

Re: [petsc-dev] subrepositories
by Jed Brown 24 Mar '11

24 Mar '11

Barry, could you explain what is so wrong about subrepos and what semantics you desire? On Fri, Mar 25, 2011 at 00:02, Barry Smith <bsmith(a)mcs.anl.gov> wrote: > > > Satish explained how hg sub repositories work. They are moronic, we don't > want to use them. > > > Is there any way with "triggers" or some other technique to rig it so if > I do hg xxx in petsc-dev it automatically runs it also in BuildSystem. For > example hg pull triggers a pull in both petsc-dev and BuildSystem? > > Barry > >

1 0

subrepositories
by Barry Smith 24 Mar '11

24 Mar '11

Satish explained how hg sub repositories work. They are moronic, we don't want to use them. Is there any way with "triggers" or some other technique to rig it so if I do hg xxx in petsc-dev it automatically runs it also in BuildSystem. For example hg pull triggers a pull in both petsc-dev and BuildSystem? Barry

1 0

Re: [petsc-dev] Updated src/sys/examples/tutorials/ex5.c
by Barry Smith 24 Mar '11

24 Mar '11

Rich, The order you call the register routines should not matter. Only the order they are listed in the struct matters. Did you get something breaking if you didn't call the register routines in order? I know the ascii view prints "incorrect" values. This is a universal problem with working with quad precision. C provides no portable (or even nonportable) way of handling format statements for different precisions builds. Unless Jed figures out a way to do it the current model is that with quad they are printed wrong and you need to manage their display yourself. Barry I try to handle the problem with %G format that we manually map to different "standard" C formats, but quad doesn't have a "standard" C format that works. On Mar 24, 2011, at 12:30 PM, Richard Katz wrote: > Attached is a diff file updating > src/sys/examples/tutorials/ex5.c > which is an example of PetscBag use. > > I've changed the order of the parameter registrations to put the Real values first, as required with quad precision. > > If you run the new version with quad precision, you'll see that when the bag is viewed to stdout, the Real numerical values are incorrectly displayed. > > Cheers > Rich > > > <ex5.diff>

1 0

Updated src/sys/examples/tutorials/ex5.c
by Richard Katz 24 Mar '11

24 Mar '11

Attached is a diff file updating src/sys/examples/tutorials/ex5.c which is an example of PetscBag use. I've changed the order of the parameter registrations to put the Real values first, as required with quad precision. If you run the new version with quad precision, you'll see that when the bag is viewed to stdout, the Real numerical values are incorrectly displayed. Cheers Rich

1 0

Re: [petsc-dev] MatNullSpaceDestroy
by Lisandro Dalcin 24 Mar '11

24 Mar '11

On 24 March 2011 13:43, Barry Smith <bsmith(a)mcs.anl.gov> wrote: > >> o now MatNullSpace instances cannot be freed with >> PetscObjectDestroy()... This change broke bad petsc4py Python+PETSc >> refcounting management. BTW, PetscObjectDestroy() does not take a >> pointer... And IMHO PetscHeaderDestroy should also take a pointer... > > Thank you for the explanation. petsc4py relies on PetscObjectDestroy() which won't work when some destroys take & and some don't. I didn't know you needed PetscObjectDestroy() to always work. > Well, long ago you said "this is polymorphism ..." :-) > > The correction solution is to switch to using the & in all destroys. It sounds like this has to happen sooner rather than later. > OK. Thanks! -- Lisandro Dalcin --------------- CIMEC (INTEC/CONICET-UNL) Predio CONICET-Santa Fe Colectora RN 168 Km 472, Paraje El Pozo 3000 Santa Fe, Argentina Tel: +54-342-4511594 (ext 1011) Tel/Fax: +54-342-4511169

1 0

Re: [petsc-dev] MatNullSpaceDestroy
by Barry Smith 24 Mar '11

24 Mar '11

> o now MatNullSpace instances cannot be freed with > PetscObjectDestroy()... This change broke bad petsc4py Python+PETSc > refcounting management. BTW, PetscObjectDestroy() does not take a > pointer... And IMHO PetscHeaderDestroy should also take a pointer... Thank you for the explanation. petsc4py relies on PetscObjectDestroy() which won't work when some destroys take & and some don't. I didn't know you needed PetscObjectDestroy() to always work. The correction solution is to switch to using the & in all destroys. It sounds like this has to happen sooner rather than later. Barry On Mar 24, 2011, at 11:28 AM, Lisandro Dalcin wrote: > On 23 March 2011 20:09, Barry Smith <bsmith(a)mcs.anl.gov> wrote: >> >> On Mar 23, 2011, at 5:58 PM, Lisandro Dalcin wrote: >> >>> Are you going to rename MatNullSpaceDestroy -> MatNullSpaceDestroy_ >>> and add the backward compatibility #define? Changeset 1a6f5fc1c27b >>> broke petsc4py... >> >> No, it was changed to the &object paradigm. >> >> Basically what is happening now is that commonly used destroys are changed with the _ paradigm >> >> Uncommonly used destroy are fixed to use the &object paradigm. >> >> Don't like the fact that there are two paradigms well I could change everything to & immediately :-( >> > > Barry, whatever you want to do, please be consistent :-)... The lines > below are from MatNullSpaceCrete > > /* cannot use MatNullSpaceDestroy() as generic destroy function > because it takes pointer to the object */ > ierr = PetscHeaderCreate(sp,_p_MatNullSpace,int,MAT_NULLSPACE_CLASSID,0,"MatNullSpace",comm,0,0);CHKERRQ(ierr); > > So now MatNullSpace instances cannot be freed with > PetscObjectDestroy()... This change broke bad petsc4py Python+PETSc > refcounting management. BTW, PetscObjectDestroy() does not take a > pointer... And IMHO PetscHeaderDestroy should also take a pointer... > > Barry, please, just reconsider this and make MatNullSpaceDestroy > follow the same pattern as Object, IS, Vec, Mat, etc. > > PS: I still think all this is going to be a mess... now one commented > about my previous proposal about a compatibility header. what's wrong > with that? > > -- > Lisandro Dalcin > --------------- > CIMEC (INTEC/CONICET-UNL) > Predio CONICET-Santa Fe > Colectora RN 168 Km 472, Paraje El Pozo > 3000 Santa Fe, Argentina > Tel: +54-342-4511594 (ext 1011) > Tel/Fax: +54-342-4511169

1 0

Re: [petsc-dev] MatNullSpaceDestroy
by Lisandro Dalcin 24 Mar '11

24 Mar '11

On 23 March 2011 20:09, Barry Smith <bsmith(a)mcs.anl.gov> wrote: > > On Mar 23, 2011, at 5:58 PM, Lisandro Dalcin wrote: > >> Are you going to rename MatNullSpaceDestroy -> MatNullSpaceDestroy_ >> and add the backward compatibility #define? Changeset 1a6f5fc1c27b >> broke petsc4py... > > No, it was changed to the &object paradigm. > > Basically what is happening now is that commonly used destroys are changed with the _ paradigm > > Uncommonly used destroy are fixed to use the &object paradigm. > > Don't like the fact that there are two paradigms well I could change everything to & immediately :-( > Barry, whatever you want to do, please be consistent :-)... The lines below are from MatNullSpaceCrete /* cannot use MatNullSpaceDestroy() as generic destroy function because it takes pointer to the object */ ierr = PetscHeaderCreate(sp,_p_MatNullSpace,int,MAT_NULLSPACE_CLASSID,0,"MatNullSpace",comm,0,0);CHKERRQ(ierr); So now MatNullSpace instances cannot be freed with PetscObjectDestroy()... This change broke bad petsc4py Python+PETSc refcounting management. BTW, PetscObjectDestroy() does not take a pointer... And IMHO PetscHeaderDestroy should also take a pointer... Barry, please, just reconsider this and make MatNullSpaceDestroy follow the same pattern as Object, IS, Vec, Mat, etc. PS: I still think all this is going to be a mess... now one commented about my previous proposal about a compatibility header. what's wrong with that? -- Lisandro Dalcin --------------- CIMEC (INTEC/CONICET-UNL) Predio CONICET-Santa Fe Colectora RN 168 Km 472, Paraje El Pozo 3000 Santa Fe, Argentina Tel: +54-342-4511594 (ext 1011) Tel/Fax: +54-342-4511169

1 0

Re: [petsc-dev] hg diff question
by Jed Brown 23 Mar '11

23 Mar '11

On Wed, Mar 23, 2011 at 23:21, Barry Smith <bsmith(a)mcs.anl.gov> wrote: > hg diff --last src/makefile hg log -p src/makefile If you haven't done it already, definitely enable the pager and color extensions by editing ~/.hgrc: [extensions] pager = color = [pager] pager = less Inside Emacs, I frequently use "M-x xhg-ann" and "M-x xhg-log" from http://download.gna.org/dvc/

1 0

Re: [petsc-dev] hg diff question
by Satish Balay 23 Mar '11

23 Mar '11

I should add: sometimes its easier to navigate the web interace of mercurial. http://petsc.cs.iit.edu/petsc/petsc-dev/ -> files http://petsc.cs.iit.edu/petsc/petsc-dev/file/5a3230381b29 -> src -> makefile revisions http://petsc.cs.iit.edu/petsc/petsc-dev/log/5a3230381b29/src/makefile -> changeset1 diff http://petsc.cs.iit.edu/petsc/petsc-dev/rev/88a383b52fbf -> [back] and changeset2 diff http://petsc.cs.iit.edu/petsc/petsc-dev/diff/2aec9ed184d1/src/makefile Satish On Wed, 23 Mar 2011, Barry Smith wrote: > > > On Mar 23, 2011, at 7:03 PM, Satish Balay wrote: > > > On Wed, 23 Mar 2011, Barry Smith wrote: > > > >> > >> On Mar 23, 2011, at 5:20 PM, Satish Balay wrote: > >> > >>> On Wed, 23 Mar 2011, Satish Balay wrote: > >>> > >>>> > >>>> Well this is what I have: > >>>> > >>>> function bsmithdiff { > >>>> hg diff -c `hg log -l 1 --template '{node}' $1` > >>>> } > >>>> > >>>> bsmithdiff src/makefile > >>> > >>> This one [bash function] is slightly better > >>> > >>> function bsmithdiff { > >>> hg diff -c `hg log -l $1 --template '{node}\n' $2| tail -1` $2 > >>> } > >>> > >>> bsmithdiff 1 src/makefile > >>> bsmithdiff 2 src/makefile > >>> > >> > >> What good is this? It doesn't show any changes. I want to see the last two sets of changes. > > > > > > It shows the last 2 changes to src/makefile for me. > > Ok, you just sent me nothing so I assumed it showed nothing. > > Barry > > > > > > > asterix:/home/balay/spetsc>function bsmithdiff { > >> hg diff -c `hg log -l $1 --template '{node}\n' $2| tail -1` $2 > >> } > > asterix:/home/balay/spetsc>bsmithdiff 1 src/makefile > > diff -r 3b37fb452c93 -r 88a383b52fbf src/makefile > > --- a/src/makefile Wed Mar 23 13:06:49 2011 +0100 > > +++ b/src/makefile Wed Mar 23 13:49:05 2011 +0100 > > @@ -1,5 +1,5 @@ > > > > -DIRS = inline sys vec mat dm ksp snes ts contrib benchmarks docs tops characteristic > > +DIRS = inline sys vec mat dm ksp snes ts contrib benchmarks docs tops > > LOCDIR = src/ > > MANSEC = Sys > > > > asterix:/home/balay/spetsc>bsmithdiff 2 src/makefile > > diff -r 2d1785e2f9cd -r 2aec9ed184d1 src/makefile > > --- a/src/makefile Mon Jun 14 17:32:56 2010 -0500 > > +++ b/src/makefile Mon Jun 14 17:51:48 2010 -0500 > > @@ -1,5 +1,5 @@ > > > > -DIRS = inline sys vec mat ksp snes ts dm contrib benchmarks docs tops characteristic > > +DIRS = inline sys vec mat dm ksp snes ts contrib benchmarks docs tops characteristic > > LOCDIR = src/ > > MANSEC = Sys > > > > asterix:/home/balay/spetsc> > > > > > > > > > > > >

1 0

Re: [petsc-dev] hg diff question
by Barry Smith 23 Mar '11

23 Mar '11

On Mar 23, 2011, at 7:03 PM, Satish Balay wrote: > On Wed, 23 Mar 2011, Barry Smith wrote: > >> >> On Mar 23, 2011, at 5:20 PM, Satish Balay wrote: >> >>> On Wed, 23 Mar 2011, Satish Balay wrote: >>> >>>> >>>> Well this is what I have: >>>> >>>> function bsmithdiff { >>>> hg diff -c `hg log -l 1 --template '{node}' $1` >>>> } >>>> >>>> bsmithdiff src/makefile >>> >>> This one [bash function] is slightly better >>> >>> function bsmithdiff { >>> hg diff -c `hg log -l $1 --template '{node}\n' $2| tail -1` $2 >>> } >>> >>> bsmithdiff 1 src/makefile >>> bsmithdiff 2 src/makefile >>> >> >> What good is this? It doesn't show any changes. I want to see the last two sets of changes. > > > It shows the last 2 changes to src/makefile for me. Ok, you just sent me nothing so I assumed it showed nothing. Barry > > > asterix:/home/balay/spetsc>function bsmithdiff { >> hg diff -c `hg log -l $1 --template '{node}\n' $2| tail -1` $2 >> } > asterix:/home/balay/spetsc>bsmithdiff 1 src/makefile > diff -r 3b37fb452c93 -r 88a383b52fbf src/makefile > --- a/src/makefile Wed Mar 23 13:06:49 2011 +0100 > +++ b/src/makefile Wed Mar 23 13:49:05 2011 +0100 > @@ -1,5 +1,5 @@ > > -DIRS = inline sys vec mat dm ksp snes ts contrib benchmarks docs tops characteristic > +DIRS = inline sys vec mat dm ksp snes ts contrib benchmarks docs tops > LOCDIR = src/ > MANSEC = Sys > > asterix:/home/balay/spetsc>bsmithdiff 2 src/makefile > diff -r 2d1785e2f9cd -r 2aec9ed184d1 src/makefile > --- a/src/makefile Mon Jun 14 17:32:56 2010 -0500 > +++ b/src/makefile Mon Jun 14 17:51:48 2010 -0500 > @@ -1,5 +1,5 @@ > > -DIRS = inline sys vec mat ksp snes ts dm contrib benchmarks docs tops characteristic > +DIRS = inline sys vec mat dm ksp snes ts contrib benchmarks docs tops characteristic > LOCDIR = src/ > MANSEC = Sys > > asterix:/home/balay/spetsc> > > > > >

1 0