Merge pull request #3057 from RosettaCommons/vmullig/fix_integration_test_using_score12
Switching rotamer_recovery integration test back to score12.
The rotamer_recovery integration test used to use the -correct flag. I took this out when I added an error message / failure for using -correct with talaris or later, since the test was not set to use a pre-talaris scorefunction (i.e. it was using ref2015 with -correct). However, this has revealed aberrant behaviour that was being masked by the spurious use of the -correct flag. Since the test is clearly testing something that's not supported post-score12, and since I don't know the original intent of the test, I'm switching it back to use score12 for now.
@everyday847 @aleaverfay
notify author
notify list [rosetta-logs@googlegroups.com]
Merge pull request #3045 from RosettaCommons/aleaverfay/fix_jd3_out_path_all
output path handling for structures written by JD3
Pose outputters for JD3 now respect the -out::path, -out::path::all, and -out::path::pdb flags.
Addressing issue brought up by @BYachnin: https://github.com/RosettaCommons/main/issues/3034
notify author
notify list [rosetta-logs@googlegroups.com]
Merge pull request #3049 from RosettaCommons/everyday847/python3_integration_tests
Changes to integration test infrastructure needed for python3 compatibility
notify author
notify list [rosetta-logs@googlegroups.com]
Merge pull request #3033 from RosettaCommons/roccomoretti/release_unit_test_fix
Fix remaining release-mode unit test issues
Minor: Change real comparison epsilon in GeneralizedEigenSolverTests to be less stringent.
Major: The Tracer set_ios_hook machinery was ostensibly insensitive to command line muting /visibility settings. However, the release-mode `operator<<` override triggers on visibility, and ignores if the ios_hook has been set. This results in a debug/release difference when `-mute all` is on.
The agreed upon fix for the moment is to change the ios_hook machinery such that it obeys the command line mute/visibility settings. That is, output from muted/invisible channels will no longer be present in the attached ios_hook stream. (However, to maintain the behavior of -mpi_tracer_to_file, the (non-command line) supermute setting will still affect only the main tracer output, not the ios_hook output.)
This change will probably not affect you unless you use the tracer redirection machinery in PyRosetta.
notify author
notify list [rosetta-logs@googlegroups.com]
Merge pull request #3040 from RosettaCommons/vmullig/fix_bluegene_runtime_issue
Fix a runtime issue on Blue Gene created by the new fconfig system
The new fconfig system is now preventing program execution on the Blue Gene/Q system. (Note: this is not @jadolfbr's fault. The problem couldn't reasonably have been anticipated before the changes went into operation.) It seems that some systems don't allow certain system calls, and this creates an unhandled exception when Rosetta starts asking for home directories and whatnot. Adding a try/catch block to handle this.
notify author
notify list [rosetta-logs@googlegroups.com]
Merge pull request #3039 from RosettaCommons/vmullig/fix_uninit_in_layerdesign
Fixing an uninitialized variable in LayerDesign.
One of the constructors failed to initialize use_symmetry_.
notify author
notify list [rosetta-logs@googlegroups.com]
Merge pull request #3032 from RosettaCommons/kalekundert/layerdesign_nataa
Make the behavior of <LayerDesign ignore_pikaa_natro="..."/> match the documentation.
notify author
notify list [rosetta-logs@googlegroups.com]
Merge pull request #3031 from RosettaCommons/JWLabonte/sugars/branching
Carbohydrates: Fix bug when glycosylating with a core glycosylated glycan
All units test pass. `KIC_with_fragments` and `rotamer_recovery` integration tests are broken. `carbohydrates` integration test change expected.
notify author
notify list [rosetta-logs@googlegroups.com]