Bugfix/fix some valgrind warnings testsuite 8 0 5

Merged
#311 · Created  · Last updated

Merged pull request

Merged in bugfix/fix_some_valgrind_warnings_testsuite_8-0-5 (pull request #311)

8b626c0·Author: ·Closed by: ·2020-06-23

Description

Replacement for #310 going into 8.0.5-rc rather than next.

  • Ensure test_fft is initialised

  • Ensure phi0 is calculated for use in final moment dumping routine in new diagnostics.

    This fixes a potential issue introduced with PR #281 which would result in this normalising quantity not being set. This is unlikely to have had any impact as the output of the called routine is probably not available due to other bugs.

  • Fix typo in gs2_io which attached an attribute to the wrong variable

  • Ensure j_ext_hist array is initialised to zero.

    Looks like there may still be a small bug with this j_ext diagnostic related to skipping istep = 1 etc.

  • Remove calls to gs2 I/O calls from new diagnostics as the new diagnostics do not set up this netcdf output

  • Remove unused gs2_io calls from new diagnostics

  • Fix use of uninitialised variable in nonlinear cyclone test

  • Fix uninitialised data when replay mode on

  • Ensure scalar results initialised in new diagnostics

    This avoids some valgrind "use of uninitialised data" errors in some of the tests. This suggests that the tests aren't appropriately calling the new diagnostics.

 

0 attachments

0 comments

Loading commits...