Merge pull request #1527 from RosettaCommons/aleaverfay/cxx11_as_default
c++11 as default
These commits flip the switch on C++11, so that we can now use its nice features (auto, lambda functions) in developing code, without having to nest that code in hard-to-read #ifdef blocks.
Lots of expected integration test changes -- a lot of low-level utility classes are now going to behave differently, and the compilers themselves are using a different standard meaning they may have to produce different code.
notify author
notify list [rosetta-logs@googlegroups.com]
Merge pull request #1588 from RosettaCommons/jadolfbr/packstat
Fix packstat score test.
So, the packstat score test was crazy with the JD2 merge (76 to >100k cycles or something). Turns out that for packstat to SCORE, it needs to write out a PDB file to the output stream to turn it into a SimplePDB. This caused the energies to fail when it tried to access the weights DURING scoring (which are now output by default).
* Use StructFileRepOptions to turn off output of the energies table (as well as any pose cache). This should fix the test.
* Use StructFileRepOptions to remove global setting of OptionSystem data Yay! @aleaverfay
Turns out that packstat is only really tested in the score performance test. If you care about packstat, you should add a proper test for it (and probably refactor it to not output an entire PDB from a pose at each score cycle).
@smlewis @roccomoretti @willsheffler
Thanks for the help in sleuthing Steven!
notify author
notify list [rosetta-logs@googlegroups.com]
Merge pull request #1586 from RosettaCommons/smlewis/mmCIFJobOutputter
Fix JD2 profile test perma-fail
Thanks @roccomoretti for pointing this out. The directories changed with the new JD2 integration tests and there was a dependent symbolic link for the profile test
@smlewis
notify author
notify list [rosetta-logs@googlegroups.com]
Merge pull request #1575 from RosettaCommons/nikoniko90/exemplar2.0
Changed exemplar generation to create more compact exemplars. Added flags to restore prior functionality, these flags will cause the make_exemplar integration test to fail. Other completely unrelated integration tests and profile tests are failing, but they look to be "usual suspects" that are failing frequently.
notify author
notify list [rosetta-logs@googlegroups.com]
Merge pull request #1534 from RosettaCommons/Cyrus/mergetarget
From Cyrus via Steven Lewis:
Four items in this PR, mostly from Steven Lewis (@smlewis), with assist from Indigo King (@indigogo):
Cyrus PR 40: smlewis/cyrus/FilterReportAsPoseExtraScoresMover; adds new class to dump Filter values mid-protocol; it seems bizarre that this didn’t exist already and/or isn’t how Filters already worked, but the list didn’t say one existed.
Cyrus PR 39: smlewis/cyrus/re_fix_beta15_catch: When I added the tool to get upset at beta_15 score functions without their requisite command line flag, it worked only for weights files with the .wts extension; now it works extensionless. Praise be to the integration tests, which caught a huge bug in the second version of this (you’re getting the third).
Cyrus PR 36: Trivial fix to the code template for Filters.
Cyrus PR 30: Updates old LoopAnalyzerMover to be RosettaScriptable. Creates LoopAnalyzerFilter, which is what it sounds like. Also Cyrus PR 40 because some of it went into a testing branch after the integration tests caught a bug, and the lack of cxx11 made me sad.
notify author
notify list [rosetta-logs@googlegroups.com]
Merge pull request #1071 from RosettaCommons/smlewis/mmCIFJobOutputter
mmCIF JD2 output compatibility - chemXRW leftover
Steven Lewis and Jared Adolf-Bryfogle
Currently, cif output is supported directly by cif_writer.hh, but not by JD2. This adds classes to enable that, plus lots of other smaller changes in extra output, dump pdb, etc.
mmCIF JD2 Options
============
Two options will now control JD2 and mmCIF:
-out:mmCIF and -out:mmCIF_gz
Output Options
============
Extra PDB and mmCIF output is now controlled through these options (listed with their current defaults):
-output_pose_energies_table true
-output_pose_cache_data true
-output_pose_fold_tree false
mmCIF Extra Tables
===============
Extra output is now refactored to be part of SFR and extracted in PoseToSFRConverter (as it should be) and can be output by the CIF writer as well. As far as we can tell, extra tables are ignored by programs, including PyMol. SO we should be in the clear. Each extra set of output is a new table for mmCIF. These are controlled by options, some of which only print out in JD2, as before, or are only output if the SFR has this info.
These new tables consist of the following:
- pose_energies
- pose_cache_data
- pose_comments
- rosetta_remarks
- rosetta_additional
The rosetta_additional table has two columns - type and output. Thats it. For now, the fold_tree is one type if we output it, and etc is another type for additional_string_output not covered by the tables above.
The Pose energies and cache table can be -not- output to a cif through options above. Alternatively, -cif_extra_data_separate_file specifies to put the scores in a separate file instead of in the cif file.
mmCIF Line Length
===============
-CIF files will have the 80 line length. It seems that the same cifparse file is used to read in and output CIF files, and it looks like you need to specify the line length in both. I'm not sure how EXTERNAL programs will respond to cif files having a maximum line length of the computer writing it out. It would be nice, but I am tabling this. What needs to happen, is that it changes with the PDB directly!
Additional Changes
==============
dump_pdb now does not have an available function where you can pass a 'tag' string, which, in reality, does absolutely nothing. This is fixed in every file that passed this bogus 'tag'.
Test Changes
==========
From what Steven and Jared can tell, we are dealing with purely cosmetic changes here for the integration tests. An extra line ending separating [extra-pdb-output] components (that were before one giant string internally).
Most deletions of lines are just lines moving around in a file. Some integration tests have their pose energies table DISAPPEAR, instead of appear - but the old state was unscored poses with empty energies tables, so that loss is still cosmetic.
Please look over your own integration tests just to double check.
Happy Wednesday!
notify author
notify list [rosetta-logs@googlegroups.com]
Merge pull request #1574 from RosettaCommons/roccomoretti/mover_status_doc
Add in-code documentation for MoverStatus codes
Also add error handling for unused MoverStatus code (MS_FAIL) in JD2.
notify author
notify list [rosetta-logs@googlegroups.com]
Merge pull request #1576 from RosettaCommons/dimaio/mg_cartmin_issue2
This PR addresses the same issue in #1566 (unfortunately, that PR did not completely fix the problem), with bondangle/bondlength minimization of Mg-containing poses. The issue arises due to the input_stub logic selecting three collinear atoms to build the stub.
This PR fixes this issue by choosing alternate atoms if the selected stub atoms are collinear.
notify author
notify list [rosetta-logs@googlegroups.com]