* Fixes#1990
makes sure we don't destroy a double bond's StereoAtoms just because a bond connected to one of its atoms is removed
* change in response to review
* spelling
* Allow types.h to be included in applications that set _USE_MATH_DEFINES
Our application defines _USE_MATH_DEFINES on the visual studio
compiler command line, so every compilation includes the macro
definition. Including types.h (or anything other RDKit header,
basically) causes compilation to fail with a "doubly defined
macro" error message.
This commit defines the macro locally and only if it is not already
defined
* remove a bunch of MSVC compiler warnings
* remove a compiler warning
* try running tests in appveyor
* typo
* investigation
* is the problem the os item?
* investigation
* output on test errors
* if only I could type
* suppress some warnings on windows (and remove a bogus attempt at doing so)
* build without serialization
* this should have been noticed for #1767
* Fixes#1810
There's still a problem with stereo atoms (see the commented out test),
but this takes care of the basics and that's up next.
* clean up that last problematic bit.
Still needs all tests to run
* this seems to be a necessary workaround for problems with VS2017.
It's a bit ugly, but shouldn't have any performance impact, so I'm not
going to get too worked up.
* This set of changes alters what we get for some of the InChI test cases.
Adjust for that.
* expose addRing to python
* backup commit
* basics of atom and bond loading
* basics of atom and bond loading
* do bond stereo too
* Loads of cleanups
Read conformers and molecule properties
Better tests
update one of the test values so that we aren't always testing ccw
* enable auto-download of rapidjson distrib
* reader now mostly done (still missing atomic/bond properties)
* a bit of optimization
* a bit of cleanup
* a bit more optimization
* backup
* test zero-order bonds
* prep work for writer (not done)
* add isotope support
* backup
* getting there with the writing
* progress on writing; still need to do the RDKit part though
* can now roundtrip, including chirality.
double bond stereo needs to wait for backend fixes
* add a timing test for benchmarking
* add a timing test for benchmarking
* add another template
* add parse params; optionally set bond types to aromatic; add python wrapper
* disable benchmark run by default
* write conformers
* refactoring
* docs
* port to commonchem
* switch representation
* start reading and writing properties
* fix a memory problem
* set a common_property for gasteiger charges
* parse partial charges
* add partial charge writing
* reformatting
* add support for disabling parts of the parsing
* remove the "name" property from files
* editing
* more post-review changes
* make the molinterchange build optional
* make them really optional
* This makes assignStereochemistry cleanIt=True not remove CIS/TRANS
bond stereo chemistry that was manually added as described in issue #1614.
Incorrect CIS/TRANS stereochemistry will still be removed by
'cleanIt=true' if symmetry is detected. However, this symmetry
detection doesn't work in more complex pseudo-stereo chemistry cases:
bond stereo that depends on other bond stereo to break symmetry; and
bond stereo that depends on other atom stereo centers to break
symmetry. Test cases for these cases have been added ifdef'd in based
on USE_NEW_STEREOCHEMISTRY.
However, getting USE_NEW_STEREOCHEMISTRY to work in a copacetic way is
not trivial, I tried a little bit here to no avail. I'm leaving the
test cases checked in as they should be useful when we decide to make
the plunge into using Canon::chiralRankMolAtoms for symmetry detection
instead of the CIP ranks.
So this fixes at least the glaring issue of STEREOCIS and STEREOTRANS
being incorrectly removed by 'cleanIt=true' when it is indeed valid
stereo. The checks made for symmetry are rudimentary, but don't feel
complete.
* add another test; make what's being tested explicit
* test smiles generation and function when Hs have been added
* add a test that fails
* I think that it's ok to remove this
* backup
* At this point all tests pass.
Bond wedging is now handled even if sanitization is turned off when mol files are read
* adjust to code changes
* fix a couple problems caused by rebase
* update docs
* Fix for issue #1730
setAromaticity() now works even if there are aromatic atoms present and the relevat test case is added
* Removed setaromaticity flag
* trying to add shape tversky
* alpha and beta no defaults, added more tverskyIndex tests to testGrid
* corrections to GridUtils.h and ShapeUtils.cpp
* corrections to GridUtils.h
* boost::thread mostly gone... still need to get rid of once
everything compiles
* replace boost::call_once
* remove link-time dependency on boost::thread
* first pass at using async
* switch to using async everywhere
* Fixes#1735
I’m unsure if this is the correct solution, it appears
empty molecules never initialize their rings.
* Return immediately when ranking empty molecules
* Make github issue the correct one
* Always resize results to the number of atoms
* first pass, does not yet actually work
* pass2, same problems
* pass2, same problems
* another test
* new tests; bugfix
* move the code out to a header
* add a double bond example
* enable auto-downloads of the code
* move the function to its own namespace
* first pass at a basic python wrapper
* change coordgen commit used
* try supporting bond stereo; does not currently work
* cis/trans seems to now work.
* first pass at templates; needs testing
* use the fixed flag too
* need mol align
* expand test
* initial pass at python wrapper for template
* simplify tests
* add an option to directly use a substructure match for alignment
* scaling
* add #define
* Define a cache setting for RDK_COORDGEN_LIBS to allow these to be used in other packages
* return the conformer id from addCoords
* Make CoordGen the default when it’s available.
This is a backup commit… the tests don’t even come close to passing.
* add some debugging options for a bit
* add alignment step to testing when using non-fixed coords
* Add global to allow use of CoordGen to be disabled
get the basic depictor tests working
* make coordgen the default when it is available
* make sure things continue to work when coordgen is disabled
* get windows builds working
* mods to get this building on windows.
something is screwy with the fileParsersTest1
* no need to generate coords for the 1K C string
* fix java wrappers
* works on linux
* update the (stupid) way dependencies were handled on windows.
this allows a lot of cleanup of cmake files (still more to do)
the linux build is unlikely to work due to the way _statics aren't handled
* docs
* extend forceRDKit applicability
* switch coordgen version
* try using templates
* try to get the template dir finding reasonably robust w.r.t. conda install
* continuing to iterate on the way the template file is installed
* fix a problem caused by the merge
* remove test that should never have been checked in
* update expected results for cartridge tests
* switch back to using the RDKit as the default coordinate generator
* Switch to uint32 for atom index to support more atoms
* Test adding and removing big number of atoms (70k)
* Change order of properites in Atom class
