Merge pull request #4587 from RosettaCommons/roccomoretti/submodule_relative
Convert .gitmodules to use relative paths.
Right now we're hard coding SSH paths for various submodules, which is annoying if people cloned main using https, especially now that we have submodules which are required for compilation.
It turns out, though, that Git can use relative paths for submodules, which allows the submodules to be cloned using the same method the main repo was.
notify author
notify list [rosetta-logs@googlegroups.com]
Merge pull request #4545 from RosettaCommons/JackMaguire/tensorflow_manager_pr1
First incremental merge of #4362
This PR includes the singleton machinery and functionality to evaluate simple networks. Networks with multiple inputs and parallel network evaluation will need to wait until a future PR.
notify author
notify list [rosetta-logs@googlegroups.com]
Merge pull request #4591 from RosettaCommons/roccomoretti/quickfix_master
Quick fixes for master tests.
Recent commits of mine broke the clang tidy and xcode tests on the server. This PR should fix them.
notify author
notify list [rosetta-logs@googlegroups.com]
Merge pull request #4475 from RosettaCommons/roccomoretti/rename_ReferenceCount
Rename ReferenceCount class to VirtualBase.
The ReferenceCount class has been a misnomer for several years now. It's purpose is no longer related to owning pointers or referencing counting, but now is simply to serve two purposes. The first is to serve as a common base class in Rosetta (such that we can store a generic ReferenceCountOP and then typesafely dynamic_cast it back to the true class), and the second is to ensure that all of the classes in a polymorphic hierarchy (i.e. one where a base class pointer/reference can point to a derived type) have a virtual destructor.
notify author
notify list [rosetta-logs@googlegroups.com]
Merge pull request #4360 from RosettaCommons/roccomoretti/boost_submod
Update Boost to 1.71, convert to submodules.
The primary goal of this PR is to update Boost to a more recent version, but as a part of it it converts Boost from a "vendorized" (code-in-main repo) to a submodule system.
The Boost submodules should be auto-updated as part of the build process (the `update_submodules.sh` script), so it should (hopefully) be invisible to most people.
Here's how it's laid out: The existing source/external/boost_1_51_0/ directory is removed, and a new source/external/boost_submod/ directory is added. (And all build systems should be updated accordingly.) Within that subdirectory are direct links to the ~86 submodules of Boost that we're currently using (from https://github.com/boostorg/). (While Boost provides a "superproject" repository, we're not using that, and as such don't have to deal with nested submodules.) The submodules should be pointing a the 1.71.0 release tag version.
Most of the submodules are being taken directly from the Boost.org's GitHub page, but there are a few with Rosetta-specific changes we're hosting directly.
The layout of the submodules doesn't quite match the layout of the released boost directory, so there's a script which will symlink the contents of the submodules into the appropriate locations in source/external/boost_submod/boost/ (it prefers to symlink directories, but there are certain things it needs to symlink as files.). Those symlinks are committed into the main repo, and shouldn't need to be regenerated unless we update Boost.
notify author
notify list [rosetta-logs@googlegroups.com]
Merge pull request #4515 from RosettaCommons/roccomoretti/defer_pose_from_parse_tag
More removal of pose usage in parse_my_tag()
There's a number of Movers, etc. which can simply have their Pose usage deferred from parse time to apply time. See the PR page for a list.
notify author
notify list [rosetta-logs@googlegroups.com]
Merge pull request #4581 from RosettaCommons/vmullig/fix_integration_tutorials_test
Fix the tutorials integration tests
Some scripts needed to be updated for Python3.
notify author
notify list [rosetta-logs@googlegroups.com]
Merge pull request #4577 from RosettaCommons/roccomoretti/repack_wo_ligand_fix
Fix RepackWithoutLigand
Amanda Loshbaugh discovered that the repack without ligand functionality was actually redesigning in some cases. This is due to the DetectProteinLigandInterface actually skipping the interface detection and just going with a default all design run in the abscence of an option. This is suboptimal, so change things such that we always detect the interface and never design in the repack without ligand case.
notify author
notify list [rosetta-logs@googlegroups.com]
Merge pull request #4579 from RosettaCommons/revert-4570-mlnance/sugar/fix_OMe_seq_on_read_in
Temporarily revert "Fixing pose building from saccharide sequence"
Reverts #4570
Pull request #4570 unfortunately caused a slowdown in pose read-in. Certain unit tests are now timing out (which may or may not be due to the pose read-in slowdown). Certain integration tests are also failing to run (e.g. test1_benchmark and dna_interface_design), which is likely one or more separate, unrelated problems. Since we don't know what all needs to be fixed, it's probably best to revert the whole pull request, fix the problems in a branch, then re-merge. This pull request therefore temporarily reverts pull request #4570 so that we can do that. @mlnance @JWLabonte @roccomoretti @everyday847
notify author
notify list [rosetta-logs@googlegroups.com]