On 16 November 2010 22:18, Barry Smith <bsmith at mcs.anl.gov> wrote: > > On Nov 16, 2010, at 5:33 PM, Lisandro Dalcin wrote: > >> On 16 November 2010 16:59, Satish Balay <balay at mcs.anl.gov> wrote: >>> We pushed a fix for this to petsc-dev. >>> >> >> So it seems that petc4py is not so broken! At least, it helped to >> discover two bugs in core PETSc!! >> >> (Sorry, Barry was so hard with me in previous mails that I could not >> resist the sarcasm) > > ?We ran your 1900 tests in 10 seconds. Running the PETSc test suite takes > like an hour and has no where near that many tests. >
However, PETSc tests do exercise the solution of actual problems, so you can catch algorithmic regressions. Also note that each test is petsc4py is actually a test for a single method, so the 1900 count is a bit artificial. And finally, let me add that many things (methods, features, etc) are still untested :-) . -- Lisandro Dalcin --------------- CIMEC (INTEC/CONICET-UNL) Predio CONICET-Santa Fe Colectora RN 168 Km 472, Paraje El Pozo Tel: +54-342-4511594 (ext 1011) Tel/Fax: +54-342-4511169
