HyTeG-HFG integration test fails
This is a list of issues detected by the integration test #1076415:
-
CMake files and various .*_all_forms.hpp
are deleted by the pipeline and thus appear in the diff -
Forms not generated by HFG -
fixed in hyteg!607 (merged)n1e1_curl_curl_affine_qe
: should have been deleted in hyteg!599 (diffs) p1_div_k_grad_affine_q1
p1_div_k_grad_blending_q1
p1_epsiloncc_blending_q2
p1_epsilonvar_blending_q2
p1_invk_mass_affine_q4
p2_epsiloncc_blending_q2
-
-
Numerous forms use a different quadrature. Apparently the quadrature selection is non-deterministic. -
The order of 2D and 3D coefficient constructor arguments is swapped. Since their data-types are the same, this looks dangerous. p0_linear_form_blending_q5
p0_linear_form_blending_q7
p1_full_stokesvar_affine_q1
p1_linear_form_blending_q5
p2_div_k_grad_affine_q4
-
The construction of Point3D
, e.g.Point3D mappedPt( {in_0, in_1, 0} );
has not-any-more-required braces. -
Some forms in HyTeG still mention quadpy in the docstring. -
Forms different in some other way p1_epsilonvar_affine_q2
p2_epsiloncc_blending_q3
There is probably more but the log reaches its maximum size at this point.
Edited by Daniel Bauer