Fwd: [petsc-maint] [spack] Petsc build with Python3
Why are we supporting this? Spack seems to generate a string of maint requests while providing marginal value at best. Matt ---------- Forwarded message --------- From: Greg Burgreen via petsc-maint <[email protected]> Date: Tue, Feb 12, 2019 at 7:53 PM Subject: Re: [petsc-maint] [spack] Petsc build with Python3 To: petsc-maint <[email protected]> Cc: [email protected] <[email protected]>, Jed Brown <[email protected]> petsc@develop worked with python-3.7.2, Satish. Thank you. The "No module named site" error did not have 'python2' in the PATH, only 'python3'. Elizabeth, approach #2 got me very close, but there were a few minor issues in that commit (for instance, the use of a deprecated string.replace in config/PETsc/Configure.py). Greg On Tue, Feb 12, 2019 at 6:34 PM Balay, Satish <[email protected]> wrote:
On Tue, 12 Feb 2019, Jed Brown via petsc-maint wrote:
Petsc4py should work fine with Python-3 (has for quite a while). The next PETSc release is due next month, give or take. You can use petsc@master in the mean time.
This is petsc@develop - in spack notation. This is fixed to work with python3.
[and petsc4py@develop should now default to using petsc@develop]
On Tue, Feb 12, 2019 at 6:28 PM Greg Burgreen < [email protected]> wrote:
I am using the latest development version of spack (commit f65a115).
The spack recipe for petsc builds fine with python-2.7-x.
When I modify the petsc package to build using python-3.x, I get "ImportError: No module named site."
I changed only one line in the petsc package to:
depends_on('[email protected]:', type='build', when='@:3.10.99')
This is a strange error. It should work [with petsc-3.10.x] if you have 'python2' in PATH.
Satish
-- 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 https://www.cse.buffalo.edu/~knepley/ <http://www.cse.buffalo.edu/~knepley/>
spack is central to xsdk project - which supports petsc. Satish On Tue, 12 Feb 2019, Matthew Knepley via petsc-dev wrote:
Why are we supporting this? Spack seems to generate a string of maint requests while providing marginal value at best.
Matt
---------- Forwarded message --------- From: Greg Burgreen via petsc-maint <[email protected]> Date: Tue, Feb 12, 2019 at 7:53 PM Subject: Re: [petsc-maint] [spack] Petsc build with Python3 To: petsc-maint <[email protected]> Cc: [email protected] <[email protected]>, Jed Brown <[email protected]>
petsc@develop worked with python-3.7.2, Satish. Thank you. The "No module named site" error did not have 'python2' in the PATH, only 'python3'.
Elizabeth, approach #2 got me very close, but there were a few minor issues in that commit (for instance, the use of a deprecated string.replace in config/PETsc/Configure.py).
Greg
On Tue, Feb 12, 2019 at 6:34 PM Balay, Satish <[email protected]> wrote:
On Tue, 12 Feb 2019, Jed Brown via petsc-maint wrote:
Petsc4py should work fine with Python-3 (has for quite a while). The next PETSc release is due next month, give or take. You can use petsc@master in the mean time.
This is petsc@develop - in spack notation. This is fixed to work with python3.
[and petsc4py@develop should now default to using petsc@develop]
On Tue, Feb 12, 2019 at 6:28 PM Greg Burgreen < [email protected]> wrote:
I am using the latest development version of spack (commit f65a115).
The spack recipe for petsc builds fine with python-2.7-x.
When I modify the petsc package to build using python-3.x, I get "ImportError: No module named site."
I changed only one line in the petsc package to:
depends_on('[email protected]:', type='build', when='@:3.10.99')
This is a strange error. It should work [with petsc-3.10.x] if you have 'python2' in PATH.
Satish
On Tue, Feb 12, 2019 at 8:10 PM Balay, Satish <[email protected]> wrote:
spack is central to xsdk project - which supports petsc.
Design mistakes are fine, but they should be corrected. Matt
Satish
On Tue, 12 Feb 2019, Matthew Knepley via petsc-dev wrote:
Why are we supporting this? Spack seems to generate a string of maint requests while providing marginal value at best.
Matt
---------- Forwarded message --------- From: Greg Burgreen via petsc-maint <[email protected]> Date: Tue, Feb 12, 2019 at 7:53 PM Subject: Re: [petsc-maint] [spack] Petsc build with Python3 To: petsc-maint <[email protected]> Cc: [email protected] <[email protected]>, Jed Brown <[email protected]>
petsc@develop worked with python-3.7.2, Satish. Thank you. The "No module named site" error did not have 'python2' in the PATH, only 'python3'.
Elizabeth, approach #2 got me very close, but there were a few minor issues in that commit (for instance, the use of a deprecated string.replace in config/PETsc/Configure.py).
Greg
On Tue, Feb 12, 2019 at 6:34 PM Balay, Satish <[email protected]> wrote:
On Tue, 12 Feb 2019, Jed Brown via petsc-maint wrote:
Petsc4py should work fine with Python-3 (has for quite a while). The next PETSc release is due next month, give or take. You can use petsc@master in the mean time.
This is petsc@develop - in spack notation. This is fixed to work with python3.
[and petsc4py@develop should now default to using petsc@develop]
On Tue, Feb 12, 2019 at 6:28 PM Greg Burgreen < [email protected]> wrote:
I am using the latest development version of spack (commit f65a115).
The spack recipe for petsc builds fine with python-2.7-x.
When I modify the petsc package to build using python-3.x, I get "ImportError: No module named site."
I changed only one line in the petsc package to:
depends_on('[email protected]:', type='build', when='@:3.10.99')
This is a strange error. It should work [with petsc-3.10.x] if you have 'python2' in PATH.
Satish
-- 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 https://www.cse.buffalo.edu/~knepley/ <http://www.cse.buffalo.edu/~knepley/>
participants (2)
-
Balay, Satish -
Matthew Knepley