Hi Roland,
I'm worried about staggered gridfunction support, any tweaks that might influence scheduling, reduction operations, and support for non-Lagrange prolongation/restrictions... just to name a few bits of carpet functionality that might not be tested by the qc0 par file.
On Wed, Oct 3, 2018, 10:25 Roland Haas rhaas@illinois.edu wrote:
Hello Zach,
I'm always nervous when a big Carpet update comes down the pipeline.
I would not call it a big update since these are changes from 2016 onwards and had they trickled in at the rate they were produced, no one would have noticed. Just lots of changes lines (due to changes of CCTK_VWarn to CCTK_WARN and reformatting of code).
Could you please confirm the full ETK testsuite passes with the new version of Carpet before pushing to master? There are a lot of Carpet features that aren't tested by qc0-mclachlan.par, but are tested by the testsuite.
Sure I will run the testsuites (must do so anyway before applying the changes), thank you for reminding me of the requirement.
Which features do you have in mind and are particularly worried about? Llama support for example is not tested in either one qc0-mclachlan or the test suites as far as I know. Note that the results for qc0-mclachlan.par are not just "similar" they are identical (in the mp_psi4_l2_m2.asc output file ie this tests quite a bit of the infrastructure).
Yours, Roland
-- My email is as private as my paper mail. I therefore support encrypting and signing email messages. Get my PGP key from http://pgp.mit.edu .