Hi,
Our goal is to release next week, at the beginning of next week if at all possible. This can only be achieved if we fix some things, and we agreed on working on it this week. Apart from an on-going discussion about include paths, we still have a long list of things to do. Since I might not have seen the progress on some points on this list I ask to report it here. What is left so far:
- hwloc 'fiasco': we need to come to a state that builds for standard systems, for everyone who tries. Right now we have conflicting reports for Ubuntu. - include path trouble on multi-arch that might (mostly) influence users that need Fortran HDF5 support. We might suggest to build HDF5 in these cases. - regenerate thorns using Kranc (Peter): status? - test gallery examples (Frank): not done yet - tutorials (Frank): will be updated once release branches exist - Queenbee (Peter): was working on it, but is not in simfactory yet - Comet (Erik): fails too many tests - Supermuc (Bruno): any update on testsuites? - Fermi & Galileo (Italian machines, Frank): testsuites all pass, simfactory entries will be added soon - mike: updated configuration exists, but was causing trouble for Erik for unknown reason that couldn't be reproduced.
At this point I would have loved to create release branches. With the current status we shouldn't. We need a little of everyone's time to make this release happen. See if you can find something to help moving things along, and if that isn't something for you, give a fresh checkout a try, to see if that works as it should on your machine.
Frank
On Fri, May 8, 2015 at 1:24 PM, Frank Loeffler knarf@cct.lsu.edu wrote:
Hi,
Our goal is to release next week, at the beginning of next week if at all possible. This can only be achieved if we fix some things, and we agreed on working on it this week. Apart from an on-going discussion about include paths, we still have a long list of things to do. Since I might not have seen the progress on some points on this list I ask to report it here. What is left so far:
- hwloc 'fiasco': we need to come to a state that builds for standard systems, for everyone who tries. Right now we have conflicting reports for Ubuntu.
let's collect real bug reports for this: complete configuration options, complete build output.
- include path trouble on multi-arch that might (mostly) influence users that need Fortran HDF5 support. We might suggest to build HDF5 in these cases.
- regenerate thorns using Kranc (Peter): status?
done
- test gallery examples (Frank): not done yet
- tutorials (Frank): will be updated once release branches exist
- Queenbee (Peter): was working on it, but is not in simfactory yet
- Comet (Erik): fails too many tests
i won't have time to investigate. comet is a new machine, so it's not a priority.
- Supermuc (Bruno): any update on testsuites?
- Fermi & Galileo (Italian machines, Frank): testsuites all pass, simfactory entries will be added soon
- mike: updated configuration exists, but was causing trouble for Erik for unknown reason that couldn't be reproduced.
i assume that the respective "softenv add" commands are missing in the "envsetup" key, or that some paths are not set manually.
At this point I would have loved to create release branches. With the
current status we shouldn't. We need a little of everyone's time to make this release happen. See if you can find something to help moving things along, and if that isn't something for you, give a fresh checkout a try, to see if that works as it should on your machine.
Frank
Users mailing list Users@einsteintoolkit.org http://lists.einsteintoolkit.org/mailman/listinfo/users
On 05/08/2015 07:24 PM, Frank Loeffler wrote:
Hi,
Our goal is to release next week, at the beginning of next week if at all possible. This can only be achieved if we fix some things, and we agreed on working on it this week. Apart from an on-going discussion about include paths, we still have a long list of things to do. Since I might not have seen the progress on some points on this list I ask to report it here. What is left so far:
- hwloc 'fiasco': we need to come to a state that builds for standard systems, for everyone who tries. Right now we have conflicting reports for Ubuntu.
- include path trouble on multi-arch that might (mostly) influence users that need Fortran HDF5 support. We might suggest to build HDF5 in these cases.
- regenerate thorns using Kranc (Peter): status?
- test gallery examples (Frank): not done yet
- tutorials (Frank): will be updated once release branches exist
- Queenbee (Peter): was working on it, but is not in simfactory yet
- Comet (Erik): fails too many tests
- Supermuc (Bruno): any update on testsuites?
I updated the test suites both for supermuc and loewe on the ET website. I don't know how to update them to the new location. Only test_local_interp_2 on one processor is failing and only on Loewe. It says the following:
WARNING[L1,P0] (TestLocalInterp2): SYMM: Error ABOVE tolerance for order 2: 4.33680869e-19 > 0.00000000e+00
- Fermi & Galileo (Italian machines, Frank): testsuites all pass, simfactory entries will be added soon
- mike: updated configuration exists, but was causing trouble for Erik for unknown reason that couldn't be reproduced.
At this point I would have loved to create release branches. With the current status we shouldn't. We need a little of everyone's time to make this release happen. See if you can find something to help moving things along, and if that isn't something for you, give a fresh checkout a try, to see if that works as it should on your machine.
Frank
Users mailing list Users@einsteintoolkit.org http://lists.einsteintoolkit.org/mailman/listinfo/users
Hello all,
I updated the test suites both for supermuc and loewe on the ET website. I don't know how to update them to the new location. Only test_local_interp_2 on one processor is failing and only on Loewe. It says the following:
WARNING[L1,P0] (TestLocalInterp2): SYMM: Error ABOVE tolerance for order 2: 4.33680869e-19 > 0.00000000e+00
There is a bug in the thorn (put in by me where I think I committed the wrong version) which causes LocalInterp2 to register the LocalInterp interpolation routines. Thus it uses the wrong interpolation routine is used. The mistake is a missing "2" in schedule.ccl.
Yours, Roland
On Fri, May 08, 2015 at 12:24:11PM -0500, Frank Loeffler wrote:
- hwloc 'fiasco': we need to come to a state that builds for standard systems, for everyone who tries. Right now we have conflicting reports for Ubuntu.
The current 'state' builds for Debian (both stable and oldstable), with a minimal package list mentioned in the option list (which installed libnuma1 and libnuma1-dev, but not specifically requested by me).
Frank
users@lists.einsteintoolkit.org
-
Bruno Coutinho Mundim -
Erik Schnetter -
Frank Loeffler -
Roland Haas