Thanks, Frank.
Wouldn't this be an argument for *removing* the data, then?
Or perhaps the test suite system could be equipped with a "short/medium/long" flag if people really wanted to avoid the most time-/memory-consuming tests?
This sounds like the kind of thing that might have been discussed before; I'm just objecting to anything other than .par files being in a "par" directory.
Bernard
On 20 November 2015 at 11:48, Frank Loeffler knarf@cct.lsu.edu wrote:
On Fri, Nov 20, 2015 at 11:46:29AM -0500, Bernard Kelly wrote:
I noticed that the thorn EinsteinInitialData/Hydro_InitExcision has a "par" subdirectory containing both parfiles *and* the resulting output directories; it looks just like a second set of test suites, except presumably they'd never be picked up by the testsuite process. Any reason why these are there (still true with new 2015_11 release)?
Without looking into it deeply I seem to remember that the problem with those was that they were taking too long, and/or using too much memory to be used as regular test suite.
Frank