#690: CarpetIOASCII modification causes tests to fail --------------------------------------+------------------------------------- Reporter: hinder | Owner: eschnett Type: defect | Status: new Priority: major | Milestone: Component: Carpet | Version: Keywords: testsuites CarpetIOASCII | --------------------------------------+------------------------------------- The following changeset,
changeset: 3377:2f2e12bdf05b tag: tip user: Erik Schnetter schnetter@cct.lsu.edu date: Tue Nov 15 11:59:53 2011 -0500 summary: CarpetIOASCII: Add new "compact" output format
causes the default CarpetIOASCII output format to change. Specifically,
{{{ # iteration 0 # refinement level 0 multigrid level 0 map 0 component 0 time level 0 # column format: 1:it 2:tl 3:rl 4:c 5:ml 6:ix 7:iy 8:iz 9:time 10:x 11:y 12:z 13:data 0 0 0 0 0 0 18 3 0 -0.473684210526316 0.0967741935483871 0 0 }}}
becomes
{{{ # iteration 0 time 0 # time level 0 # refinement level 0 multigrid level 0 map 0 component 0 # column format: 1:it 2:tl 3:rl 4:c 5:ml 6:ix 7:iy 8:iz 9:time 10:x 11:y 12:z 13:data
0 0 0 0 0 0 18 3 0 -0.473684210526316 0.0967741935483871 0 0 }}}
This causes certain test cases (I think those which do not use IO::out_fileinfo = "none") to fail. The problem is not in the modification of the comment lines, whose content is ignored by the test mechanism, but in the addition of a blank line or a comment line (the "time level 0" in the above example). The test mechanism compares files line-by-line, and so becomes out of step if additional comment or blank lines are introduced.
The attached patch modifies the test mechanism to completely ignore the presence of any blank or comment lines. With this patch, the tests bhns_eval, bhns_interp, checkpointML and recoverML now pass again.
Is this solution acceptable? Should the test mechanism be modified in a different way? Or should CarpetIOASCII be modified to produce the old format?