Re: [Rdkit-discuss] RDKit build issues
Yeah, please send me (no reason to send them to the whole list) the ctest -V logs from running those failing tests individually and we'll see if we can figure this out. -greg On Tue, Apr 19, 2016 at 10:26 PM, Matthew Lardywrote: > > I've been doing that too, and I am very happy that I've got the Java > wrappers built (v2016.03.1). I've got four Python operations that fail. I > can send the log from ctest -V if you'd like: > > 97% tests passed, 4 tests failed out of 133 > > Total Test time (real) = 464.59 sec > > The following tests FAILED: > 44 - pyChemReactions (Failed) > 62 - pyForceFieldHelpers (Failed) > 86 - pyGraphMolWrap (Failed) > 133 - pythonTestDirChem (Failed) > Errors while running CTest > > Thanks for everyone's help! > Matt > -- Find and fix application performance issues faster with Applications Manager Applications Manager provides deep performance insights into multiple tiers of your business applications. It resolves application problems quickly and reduces your MTTR. Get your free trial! https://ad.doubleclick.net/ddm/clk/302982198;130105516;z___ Rdkit-discuss mailing list Rdkit-discuss@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/rdkit-discuss
Re: [Rdkit-discuss] RDKit build issues
Hi Greg, I've been doing that too, and I am very happy that I've got the Java wrappers built (v2016.03.1). I've got four Python operations that fail. I can send the log from ctest -V if you'd like: 97% tests passed, 4 tests failed out of 133 Total Test time (real) = 464.59 sec The following tests FAILED: 44 - pyChemReactions (Failed) 62 - pyForceFieldHelpers (Failed) 86 - pyGraphMolWrap (Failed) 133 - pythonTestDirChem (Failed) Errors while running CTest Thanks for everyone's help! Matt On Tue, Apr 19, 2016 at 4:23 AM, Greg Landrumwrote: > It's great that you're almost there. > > The best way to help diagnose problems running the tests is to run ctest > in verbose mode with "-V". Here's an example on my machine: > > glandrum@Otter:/scratch/RDKit_git/build$ echo $RDBASE > /scratch/RDKit_git > glandrum@Otter:/scratch/RDKit_git/build$ echo $LD_LIBRARY_PATH > /scratch/RDKit_git/lib:/usr/local/pgsql/lib: > glandrum@Otter:/scratch/RDKit_git/build$ ctest -V -R JavaAromat > UpdateCTestConfiguration from > :/scratch/RDKit_git/build/DartConfiguration.tcl > UpdateCTestConfiguration from > :/scratch/RDKit_git/build/DartConfiguration.tcl > Test project /scratch/RDKit_git/build > Constructing a list of tests > Done constructing a list of tests > Checking test dependency graph... > Checking test dependency graph end > test 99 > Start 99: JavaAromaticTests > > 99: Test command: /usr/bin/java > "-Djava.library.path=/scratch/RDKit_git/Code/JavaWrappers/gmwrapper" "-cp" > "/scratch/RDKit_git/External/java_lib/junit.jar:/scratch/RDKit_git/Code/JavaWrappers/gmwrapper/build-test:/scratch/RDKit_git/Code/JavaWrappers/gmwrapper/org.RDKit.jar" > "org.RDKit.AromaticTests" > 99: Test timeout computed to be: 9.99988e+06 > 99: JUnit version 4.10 > 99: .. > 99: Time: 1.8 > 99: > 99: OK (2 tests) > 99: > 1/1 Test #99: JavaAromaticTests Passed1.99 sec > > The following tests passed: > JavaAromaticTests > > 100% tests passed, 0 tests failed out of 1 > > Total Test time (real) = 2.00 sec > glandrum@Otter:/scratch/RDKit_git/build$ > > > I do not have CLASSPATH set and I don't think it's necessary in order to > be able to run the tests. > > -greg > > > > > On Mon, Apr 18, 2016 at 11:47 PM, Matthew Lardy wrote: > >> Hi Brian, >> >> Thanks! I gave it a whirl, but no improvement. In previous installs >> I've specified the shared object directly, but there isn't a good reason >> for me to do so. I've done a make install, so everything should be built >> wherever it is supposed to go. >> >> I've missed something simple. I just don't see it yet, >> Matt >> >> >> On Mon, Apr 18, 2016 at 2:08 PM, Brian Kelley >> wrote: >> >>> > $RDBASE/Code/JavaWrappers/gmwrapper/libGraphMolWrap.so >>> >>> You probably need to drop the explicit .so, the library path is the path >>> to the directory not to the .so itself. >>> >>> Here is what I normally use for testing ( note I usually build into an >>> rdkit_build directory via: -DCMAKE_INSTALL_PREFIX=>> rdkit_build> ) >>> >>> >>> RDBASE=`pwd`/../../rdkit LD_LIBRARY_PATH=`pwd`/rdkit_build/lib >>> PYTHONPATH=`pwd`/rdkit_build/lib/python2.7/site-packages ctest >>> >>> On Mon, Apr 18, 2016 at 4:57 PM, Matthew Lardy wrote: >>> A quick question, what should be set up for the environment for the Java wrappers? I've got the following set: setenv RDBASE $WORKSPACE/RDKit-2016/rdkit-Release_2016_03 setenv PATH ${PATH}:$RDBASE/lib setenv CLASSPATH $RDBASE/Code/JavaWrappers/gmwrapper/org.RDKit.jar setenv LD_LIBRARY_PATH ${LD_LIBRARY_PATH}:$WORKSPACE/boost1.59/lib:$RDBASE/build/lib:$RDBASE/Code/JavaWrappers/gmwrapper/libGraphMolWrap.so setenv PYTHONPATH ${PYTHONPATH}:$RDBASE The code builds without issue, save a few warnings. Here is the Java error I am getting (during the tests): JUnit version 4.12 119: Exception in thread "main" java.lang.NoClassDefFoundError: org/hamcrest/SelfDescribing Is my CLASSPATH wrong? Did I forget a SO in my LD_LIBRARY_PATH? Thanks in advance! Matt On Mon, Apr 18, 2016 at 9:56 AM, Matthew Lardy wrote: > My thoughts exactly. I've forgotten to set RDBASE before and had a > few tests fail. This is different. It was everything and it took me days > to get to a place where some of the base code was passing tests. Someone > mentioned that I need to build boost with the version of Python that I am > using. This is problem in an environment where people choose their own > adventure with modules. I have no guarantee that they won't chose Python > 3.x when I've compiled everything with 2.7.7. That problem will be a good > to have when I get there. :) > > If my present build doesn't pass everything, and it won't, I'll send > you the fun
Re: [Rdkit-discuss] RDKit build issues
Hi Brian, Thanks! I gave it a whirl, but no improvement. In previous installs I've specified the shared object directly, but there isn't a good reason for me to do so. I've done a make install, so everything should be built wherever it is supposed to go. I've missed something simple. I just don't see it yet, Matt On Mon, Apr 18, 2016 at 2:08 PM, Brian Kelleywrote: > > $RDBASE/Code/JavaWrappers/gmwrapper/libGraphMolWrap.so > > You probably need to drop the explicit .so, the library path is the path > to the directory not to the .so itself. > > Here is what I normally use for testing ( note I usually build into an > rdkit_build directory via: -DCMAKE_INSTALL_PREFIX= > ) > > > RDBASE=`pwd`/../../rdkit LD_LIBRARY_PATH=`pwd`/rdkit_build/lib > PYTHONPATH=`pwd`/rdkit_build/lib/python2.7/site-packages ctest > > On Mon, Apr 18, 2016 at 4:57 PM, Matthew Lardy wrote: > >> A quick question, what should be set up for the environment for the Java >> wrappers? I've got the following set: >> >> setenv RDBASE $WORKSPACE/RDKit-2016/rdkit-Release_2016_03 >> setenv PATH ${PATH}:$RDBASE/lib >> >> setenv CLASSPATH $RDBASE/Code/JavaWrappers/gmwrapper/org.RDKit.jar >> setenv LD_LIBRARY_PATH >> ${LD_LIBRARY_PATH}:$WORKSPACE/boost1.59/lib:$RDBASE/build/lib:$RDBASE/Code/JavaWrappers/gmwrapper/libGraphMolWrap.so >> setenv PYTHONPATH ${PYTHONPATH}:$RDBASE >> >> The code builds without issue, save a few warnings. Here is the Java >> error I am getting (during the tests): >> >> JUnit version 4.12 >> 119: Exception in thread "main" java.lang.NoClassDefFoundError: >> org/hamcrest/SelfDescribing >> >> Is my CLASSPATH wrong? Did I forget a SO in my LD_LIBRARY_PATH? >> >> Thanks in advance! >> Matt >> >> >> >> On Mon, Apr 18, 2016 at 9:56 AM, Matthew Lardy wrote: >> >>> My thoughts exactly. I've forgotten to set RDBASE before and had a few >>> tests fail. This is different. It was everything and it took me days to >>> get to a place where some of the base code was passing tests. Someone >>> mentioned that I need to build boost with the version of Python that I am >>> using. This is problem in an environment where people choose their own >>> adventure with modules. I have no guarantee that they won't chose Python >>> 3.x when I've compiled everything with 2.7.7. That problem will be a good >>> to have when I get there. :) >>> >>> If my present build doesn't pass everything, and it won't, I'll send you >>> the fun directly. :) >>> >>> Thanks again Greg! >>> Matt >>> >>> >>> On Fri, Apr 15, 2016 at 9:34 PM, Greg Landrum >>> wrote: >>> Brian's suggestion to take a look at the cmake invocation in the travis file is a good one. Based on the number of failing tests, I think something more than just an RDBASE problem is wrong. The best way to track this down is by sending me the cmake command you ran, the output of running make, and the output of running ctest (after you run ctest you will find this in the directory Testing/Temporary/LastTest.log). Because this is pretty large and likely not of interest to others, it's probably best to send it directly to me or to create a gist and send the link to that. -greg On Sat, Apr 16, 2016 at 2:22 AM, Brian Kelley wrote: > Getting the right version of boost can be tricky. You can see our > normal cmake incantation here as well as how we set RDBASE for tests > > https://github.com/rdkit/rdkit/blob/master/.travis.yml > > Note the > > -D Boost_NO_SYSTEM_PATHS=ON > > When running cmake, otherwise cmake can get very confused. > > > > Brian Kelley > > On Apr 15, 2016, at 6:26 PM, Matthew Lardy wrote: > > I'll add, that remembering that cmake and ccmake can produce different > outcomes I've gone back to trying cmake. But I can't overwrite the > variables in cmake. Here are the results from trying to specify them: > > Manually-specified variables were not used by the project: > > BOOST_DIR > BOOST_INCLUDE_DIR > BOOST_LIBRARY_DIR > BOOST_PYTHON_LIBRARY_DEBUG > BOOST_PYTHON_LIBRARY_RELEASE > BOOST_REGEX_LIBRARY_DEBUG > BOOST_REGEX_LIBRARY_RELEASE > BOOST_SERIALIZATION_LIBRARY_DEBUG > BOOST_SERIALIZATION_LIBRARY_RELEASE > BOOST_SYSTEM_LIBRARY_DEBUG > BOOST_SYSTEM_LIBRARY_RELEASE > BOOST_THREAD_LIBRARY_DEBUG > BOOST_THREAD_LIBRARY_RELEASE > > cmake, as per it's usual, picked up an ancient version of boost (which > I want to override). I can get around this with ccmake, but nothing that > I > compile can pass all of the tests. If someone knows what these variables > (shown in ccmake) are called, I'd love to know. > > Thanks in advance! >
Re: [Rdkit-discuss] RDKit build issues
A quick question, what should be set up for the environment for the Java wrappers? I've got the following set: setenv RDBASE $WORKSPACE/RDKit-2016/rdkit-Release_2016_03 setenv PATH ${PATH}:$RDBASE/lib setenv CLASSPATH $RDBASE/Code/JavaWrappers/gmwrapper/org.RDKit.jar setenv LD_LIBRARY_PATH ${LD_LIBRARY_PATH}:$WORKSPACE/boost1.59/lib:$RDBASE/build/lib:$RDBASE/Code/JavaWrappers/gmwrapper/libGraphMolWrap.so setenv PYTHONPATH ${PYTHONPATH}:$RDBASE The code builds without issue, save a few warnings. Here is the Java error I am getting (during the tests): JUnit version 4.12 119: Exception in thread "main" java.lang.NoClassDefFoundError: org/hamcrest/SelfDescribing Is my CLASSPATH wrong? Did I forget a SO in my LD_LIBRARY_PATH? Thanks in advance! Matt On Mon, Apr 18, 2016 at 9:56 AM, Matthew Lardywrote: > My thoughts exactly. I've forgotten to set RDBASE before and had a few > tests fail. This is different. It was everything and it took me days to > get to a place where some of the base code was passing tests. Someone > mentioned that I need to build boost with the version of Python that I am > using. This is problem in an environment where people choose their own > adventure with modules. I have no guarantee that they won't chose Python > 3.x when I've compiled everything with 2.7.7. That problem will be a good > to have when I get there. :) > > If my present build doesn't pass everything, and it won't, I'll send you > the fun directly. :) > > Thanks again Greg! > Matt > > > On Fri, Apr 15, 2016 at 9:34 PM, Greg Landrum > wrote: > >> Brian's suggestion to take a look at the cmake invocation in the travis >> file is a good one. >> >> Based on the number of failing tests, I think something more than just an >> RDBASE problem is wrong. The best way to track this down is by sending me >> the cmake command you ran, the output of running make, and the output of >> running ctest (after you run ctest you will find this in the directory >> Testing/Temporary/LastTest.log). Because this is pretty large and likely >> not of interest to others, it's probably best to send it directly to me or >> to create a gist and send the link to that. >> >> -greg >> >> >> >> On Sat, Apr 16, 2016 at 2:22 AM, Brian Kelley >> wrote: >> >>> Getting the right version of boost can be tricky. You can see our >>> normal cmake incantation here as well as how we set RDBASE for tests >>> >>> https://github.com/rdkit/rdkit/blob/master/.travis.yml >>> >>> Note the >>> >>> -D Boost_NO_SYSTEM_PATHS=ON >>> >>> When running cmake, otherwise cmake can get very confused. >>> >>> >>> >>> Brian Kelley >>> >>> On Apr 15, 2016, at 6:26 PM, Matthew Lardy wrote: >>> >>> I'll add, that remembering that cmake and ccmake can produce different >>> outcomes I've gone back to trying cmake. But I can't overwrite the >>> variables in cmake. Here are the results from trying to specify them: >>> >>> Manually-specified variables were not used by the project: >>> >>> BOOST_DIR >>> BOOST_INCLUDE_DIR >>> BOOST_LIBRARY_DIR >>> BOOST_PYTHON_LIBRARY_DEBUG >>> BOOST_PYTHON_LIBRARY_RELEASE >>> BOOST_REGEX_LIBRARY_DEBUG >>> BOOST_REGEX_LIBRARY_RELEASE >>> BOOST_SERIALIZATION_LIBRARY_DEBUG >>> BOOST_SERIALIZATION_LIBRARY_RELEASE >>> BOOST_SYSTEM_LIBRARY_DEBUG >>> BOOST_SYSTEM_LIBRARY_RELEASE >>> BOOST_THREAD_LIBRARY_DEBUG >>> BOOST_THREAD_LIBRARY_RELEASE >>> >>> cmake, as per it's usual, picked up an ancient version of boost (which I >>> want to override). I can get around this with ccmake, but nothing that I >>> compile can pass all of the tests. If someone knows what these variables >>> (shown in ccmake) are called, I'd love to know. >>> >>> Thanks in advance! >>> Matthew >>> >>> >>> On Fri, Apr 15, 2016 at 2:58 PM, Matthew Lardy wrote: >>> Hi all, If someone has an insight I would love to hear it about how best to build RDKit from scratch. I am using the following to build the 2015.03 release: RedHat Linux version 6.4 gcc version 4.4.7 20120313 (Red Hat 4.4.7-11) (GCC) swig version 3.0.8 cmake version 3.0.0 boost version 1.53.0 (I've also tried boost 1.59.0) python version 2.7.7 I use ccmake and alter the things which need to be altered (as cmake always grabs the wrong things for me). My build compiles nicely, and without errors. But the tests are a mess. Here is the summary of my build: 26% tests passed, 87 tests failed out of 118 I've put the details of the tests, in the hope that someone sees a pattern that I do not, below. Before someone recommends a clean install that is not possible. I also do not have root, so I can't just wipe the machine and start over. I cannot use an RPM, so that is out too. Any ideas are welcome! Thanks, Matthew
Re: [Rdkit-discuss] RDKit build issues
My thoughts exactly. I've forgotten to set RDBASE before and had a few tests fail. This is different. It was everything and it took me days to get to a place where some of the base code was passing tests. Someone mentioned that I need to build boost with the version of Python that I am using. This is problem in an environment where people choose their own adventure with modules. I have no guarantee that they won't chose Python 3.x when I've compiled everything with 2.7.7. That problem will be a good to have when I get there. :) If my present build doesn't pass everything, and it won't, I'll send you the fun directly. :) Thanks again Greg! Matt On Fri, Apr 15, 2016 at 9:34 PM, Greg Landrumwrote: > Brian's suggestion to take a look at the cmake invocation in the travis > file is a good one. > > Based on the number of failing tests, I think something more than just an > RDBASE problem is wrong. The best way to track this down is by sending me > the cmake command you ran, the output of running make, and the output of > running ctest (after you run ctest you will find this in the directory > Testing/Temporary/LastTest.log). Because this is pretty large and likely > not of interest to others, it's probably best to send it directly to me or > to create a gist and send the link to that. > > -greg > > > > On Sat, Apr 16, 2016 at 2:22 AM, Brian Kelley > wrote: > >> Getting the right version of boost can be tricky. You can see our normal >> cmake incantation here as well as how we set RDBASE for tests >> >> https://github.com/rdkit/rdkit/blob/master/.travis.yml >> >> Note the >> >> -D Boost_NO_SYSTEM_PATHS=ON >> >> When running cmake, otherwise cmake can get very confused. >> >> >> >> Brian Kelley >> >> On Apr 15, 2016, at 6:26 PM, Matthew Lardy wrote: >> >> I'll add, that remembering that cmake and ccmake can produce different >> outcomes I've gone back to trying cmake. But I can't overwrite the >> variables in cmake. Here are the results from trying to specify them: >> >> Manually-specified variables were not used by the project: >> >> BOOST_DIR >> BOOST_INCLUDE_DIR >> BOOST_LIBRARY_DIR >> BOOST_PYTHON_LIBRARY_DEBUG >> BOOST_PYTHON_LIBRARY_RELEASE >> BOOST_REGEX_LIBRARY_DEBUG >> BOOST_REGEX_LIBRARY_RELEASE >> BOOST_SERIALIZATION_LIBRARY_DEBUG >> BOOST_SERIALIZATION_LIBRARY_RELEASE >> BOOST_SYSTEM_LIBRARY_DEBUG >> BOOST_SYSTEM_LIBRARY_RELEASE >> BOOST_THREAD_LIBRARY_DEBUG >> BOOST_THREAD_LIBRARY_RELEASE >> >> cmake, as per it's usual, picked up an ancient version of boost (which I >> want to override). I can get around this with ccmake, but nothing that I >> compile can pass all of the tests. If someone knows what these variables >> (shown in ccmake) are called, I'd love to know. >> >> Thanks in advance! >> Matthew >> >> >> On Fri, Apr 15, 2016 at 2:58 PM, Matthew Lardy wrote: >> >>> Hi all, >>> >>> If someone has an insight I would love to hear it about how best to >>> build RDKit from scratch. >>> >>> I am using the following to build the 2015.03 release: >>> RedHat Linux version 6.4 >>> gcc version 4.4.7 20120313 (Red Hat 4.4.7-11) (GCC) >>> swig version 3.0.8 >>> cmake version 3.0.0 >>> boost version 1.53.0 (I've also tried boost 1.59.0) >>> python version 2.7.7 >>> >>> I use ccmake and alter the things which need to be altered (as cmake >>> always grabs the wrong things for me). My build compiles nicely, and >>> without errors. But the tests are a mess. Here is the summary of my build: >>> >>> 26% tests passed, 87 tests failed out of 118 >>> >>> I've put the details of the tests, in the hope that someone sees a >>> pattern that I do not, below. >>> >>> Before someone recommends a clean install that is not possible. I also >>> do not have root, so I can't just wipe the machine and start over. I >>> cannot use an RPM, so that is out too. Any ideas are welcome! >>> >>> Thanks, >>> Matthew >>> >>> The following tests FAILED: >>> 2 - testDataStructs (OTHER_FAULT) >>> 3 - pyBV (Failed) >>> 4 - pyDiscreteValueVect (Failed) >>> 5 - pySparseIntVect (Failed) >>> 7 - testGrid (OTHER_FAULT) >>> 8 - testPyGeometry (Failed) >>> 11 - pyAlignment (Failed) >>> 14 - testMMFFForceField (OTHER_FAULT) >>> 15 - pyForceFieldConstraints (Failed) >>> 17 - pyDistGeom (Failed) >>> 18 - graphmolTest1 (OTHER_FAULT) >>> 21 - graphmolMolOpsTest (SEGFAULT) >>> 23 - graphmoltestChirality (OTHER_FAULT) >>> 24 - graphmoltestPickler (OTHER_FAULT) >>> 26 - hanoiTest (OTHER_FAULT) >>> 28 - testDepictor (OTHER_FAULT) >>> 29 - pyDepictor (Failed) >>> 32 - fileParsersTest1 (OTHER_FAULT) >>> 33 - testMolSupplier (OTHER_FAULT) >>> 34 - testMolWriter (OTHER_FAULT) >>> 35 - testTplParser
Re: [Rdkit-discuss] RDKit build issues
Hi Brian, The RDBASE was pointed out to me by others as well. It appears that rebuilding with cmake and forcing things via the command-line is working out better than it did earlier. I'm now up to just the python things failing (although I haven't attempted to build the Java wrappers yet). I am, clearly, all thumbs with cmake if things don't go well. :) Worse is that in my environment nothing is stock. In fact, I have to go back and rebuild boost as it also found the ancient libraries that it shouldn't be touching. The C++ code is compiling and passing tests now. No more failures! Thank you all!! Matt On Fri, Apr 15, 2016 at 5:22 PM, Brian Kelleywrote: > Getting the right version of boost can be tricky. You can see our normal > cmake incantation here as well as how we set RDBASE for tests > > https://github.com/rdkit/rdkit/blob/master/.travis.yml > > Note the > > -D Boost_NO_SYSTEM_PATHS=ON > > When running cmake, otherwise cmake can get very confused. > > > > Brian Kelley > > On Apr 15, 2016, at 6:26 PM, Matthew Lardy wrote: > > I'll add, that remembering that cmake and ccmake can produce different > outcomes I've gone back to trying cmake. But I can't overwrite the > variables in cmake. Here are the results from trying to specify them: > > Manually-specified variables were not used by the project: > > BOOST_DIR > BOOST_INCLUDE_DIR > BOOST_LIBRARY_DIR > BOOST_PYTHON_LIBRARY_DEBUG > BOOST_PYTHON_LIBRARY_RELEASE > BOOST_REGEX_LIBRARY_DEBUG > BOOST_REGEX_LIBRARY_RELEASE > BOOST_SERIALIZATION_LIBRARY_DEBUG > BOOST_SERIALIZATION_LIBRARY_RELEASE > BOOST_SYSTEM_LIBRARY_DEBUG > BOOST_SYSTEM_LIBRARY_RELEASE > BOOST_THREAD_LIBRARY_DEBUG > BOOST_THREAD_LIBRARY_RELEASE > > cmake, as per it's usual, picked up an ancient version of boost (which I > want to override). I can get around this with ccmake, but nothing that I > compile can pass all of the tests. If someone knows what these variables > (shown in ccmake) are called, I'd love to know. > > Thanks in advance! > Matthew > > > On Fri, Apr 15, 2016 at 2:58 PM, Matthew Lardy wrote: > >> Hi all, >> >> If someone has an insight I would love to hear it about how best to build >> RDKit from scratch. >> >> I am using the following to build the 2015.03 release: >> RedHat Linux version 6.4 >> gcc version 4.4.7 20120313 (Red Hat 4.4.7-11) (GCC) >> swig version 3.0.8 >> cmake version 3.0.0 >> boost version 1.53.0 (I've also tried boost 1.59.0) >> python version 2.7.7 >> >> I use ccmake and alter the things which need to be altered (as cmake >> always grabs the wrong things for me). My build compiles nicely, and >> without errors. But the tests are a mess. Here is the summary of my build: >> >> 26% tests passed, 87 tests failed out of 118 >> >> I've put the details of the tests, in the hope that someone sees a >> pattern that I do not, below. >> >> Before someone recommends a clean install that is not possible. I also >> do not have root, so I can't just wipe the machine and start over. I >> cannot use an RPM, so that is out too. Any ideas are welcome! >> >> Thanks, >> Matthew >> >> The following tests FAILED: >> 2 - testDataStructs (OTHER_FAULT) >> 3 - pyBV (Failed) >> 4 - pyDiscreteValueVect (Failed) >> 5 - pySparseIntVect (Failed) >> 7 - testGrid (OTHER_FAULT) >> 8 - testPyGeometry (Failed) >> 11 - pyAlignment (Failed) >> 14 - testMMFFForceField (OTHER_FAULT) >> 15 - pyForceFieldConstraints (Failed) >> 17 - pyDistGeom (Failed) >> 18 - graphmolTest1 (OTHER_FAULT) >> 21 - graphmolMolOpsTest (SEGFAULT) >> 23 - graphmoltestChirality (OTHER_FAULT) >> 24 - graphmoltestPickler (OTHER_FAULT) >> 26 - hanoiTest (OTHER_FAULT) >> 28 - testDepictor (OTHER_FAULT) >> 29 - pyDepictor (Failed) >> 32 - fileParsersTest1 (OTHER_FAULT) >> 33 - testMolSupplier (OTHER_FAULT) >> 34 - testMolWriter (OTHER_FAULT) >> 35 - testTplParser (OTHER_FAULT) >> 36 - testMol2ToMol (OTHER_FAULT) >> 38 - testReaction (OTHER_FAULT) >> 40 - pyChemReactions (Failed) >> 41 - testChemTransforms (OTHER_FAULT) >> 44 - testFragCatalog (OTHER_FAULT) >> 45 - pyFragCatalog (Failed) >> 46 - testDescriptors (OTHER_FAULT) >> 47 - pyMolDescriptors (Failed) >> 48 - testFingerprints (OTHER_FAULT) >> 50 - pyPartialCharges (Failed) >> 51 - testMolTransforms (OTHER_FAULT) >> 52 - pyMolTransforms (Failed) >> 53 - testMMFFForceFieldHelpers (OTHER_FAULT) >> 54 - testUFFForceFieldHelpers (OTHER_FAULT) >> 55 - pyForceFieldHelpers (Failed) >> 56 - testDistGeomHelpers (OTHER_FAULT) >> 57 - pyDistGeom (Failed) >> 58 - testMolAlign
Re: [Rdkit-discuss] RDKit build issues
Brian's suggestion to take a look at the cmake invocation in the travis file is a good one. Based on the number of failing tests, I think something more than just an RDBASE problem is wrong. The best way to track this down is by sending me the cmake command you ran, the output of running make, and the output of running ctest (after you run ctest you will find this in the directory Testing/Temporary/LastTest.log). Because this is pretty large and likely not of interest to others, it's probably best to send it directly to me or to create a gist and send the link to that. -greg On Sat, Apr 16, 2016 at 2:22 AM, Brian Kelleywrote: > Getting the right version of boost can be tricky. You can see our normal > cmake incantation here as well as how we set RDBASE for tests > > https://github.com/rdkit/rdkit/blob/master/.travis.yml > > Note the > > -D Boost_NO_SYSTEM_PATHS=ON > > When running cmake, otherwise cmake can get very confused. > > > > Brian Kelley > > On Apr 15, 2016, at 6:26 PM, Matthew Lardy wrote: > > I'll add, that remembering that cmake and ccmake can produce different > outcomes I've gone back to trying cmake. But I can't overwrite the > variables in cmake. Here are the results from trying to specify them: > > Manually-specified variables were not used by the project: > > BOOST_DIR > BOOST_INCLUDE_DIR > BOOST_LIBRARY_DIR > BOOST_PYTHON_LIBRARY_DEBUG > BOOST_PYTHON_LIBRARY_RELEASE > BOOST_REGEX_LIBRARY_DEBUG > BOOST_REGEX_LIBRARY_RELEASE > BOOST_SERIALIZATION_LIBRARY_DEBUG > BOOST_SERIALIZATION_LIBRARY_RELEASE > BOOST_SYSTEM_LIBRARY_DEBUG > BOOST_SYSTEM_LIBRARY_RELEASE > BOOST_THREAD_LIBRARY_DEBUG > BOOST_THREAD_LIBRARY_RELEASE > > cmake, as per it's usual, picked up an ancient version of boost (which I > want to override). I can get around this with ccmake, but nothing that I > compile can pass all of the tests. If someone knows what these variables > (shown in ccmake) are called, I'd love to know. > > Thanks in advance! > Matthew > > > On Fri, Apr 15, 2016 at 2:58 PM, Matthew Lardy wrote: > >> Hi all, >> >> If someone has an insight I would love to hear it about how best to build >> RDKit from scratch. >> >> I am using the following to build the 2015.03 release: >> RedHat Linux version 6.4 >> gcc version 4.4.7 20120313 (Red Hat 4.4.7-11) (GCC) >> swig version 3.0.8 >> cmake version 3.0.0 >> boost version 1.53.0 (I've also tried boost 1.59.0) >> python version 2.7.7 >> >> I use ccmake and alter the things which need to be altered (as cmake >> always grabs the wrong things for me). My build compiles nicely, and >> without errors. But the tests are a mess. Here is the summary of my build: >> >> 26% tests passed, 87 tests failed out of 118 >> >> I've put the details of the tests, in the hope that someone sees a >> pattern that I do not, below. >> >> Before someone recommends a clean install that is not possible. I also >> do not have root, so I can't just wipe the machine and start over. I >> cannot use an RPM, so that is out too. Any ideas are welcome! >> >> Thanks, >> Matthew >> >> The following tests FAILED: >> 2 - testDataStructs (OTHER_FAULT) >> 3 - pyBV (Failed) >> 4 - pyDiscreteValueVect (Failed) >> 5 - pySparseIntVect (Failed) >> 7 - testGrid (OTHER_FAULT) >> 8 - testPyGeometry (Failed) >> 11 - pyAlignment (Failed) >> 14 - testMMFFForceField (OTHER_FAULT) >> 15 - pyForceFieldConstraints (Failed) >> 17 - pyDistGeom (Failed) >> 18 - graphmolTest1 (OTHER_FAULT) >> 21 - graphmolMolOpsTest (SEGFAULT) >> 23 - graphmoltestChirality (OTHER_FAULT) >> 24 - graphmoltestPickler (OTHER_FAULT) >> 26 - hanoiTest (OTHER_FAULT) >> 28 - testDepictor (OTHER_FAULT) >> 29 - pyDepictor (Failed) >> 32 - fileParsersTest1 (OTHER_FAULT) >> 33 - testMolSupplier (OTHER_FAULT) >> 34 - testMolWriter (OTHER_FAULT) >> 35 - testTplParser (OTHER_FAULT) >> 36 - testMol2ToMol (OTHER_FAULT) >> 38 - testReaction (OTHER_FAULT) >> 40 - pyChemReactions (Failed) >> 41 - testChemTransforms (OTHER_FAULT) >> 44 - testFragCatalog (OTHER_FAULT) >> 45 - pyFragCatalog (Failed) >> 46 - testDescriptors (OTHER_FAULT) >> 47 - pyMolDescriptors (Failed) >> 48 - testFingerprints (OTHER_FAULT) >> 50 - pyPartialCharges (Failed) >> 51 - testMolTransforms (OTHER_FAULT) >> 52 - pyMolTransforms (Failed) >> 53 - testMMFFForceFieldHelpers (OTHER_FAULT) >> 54 - testUFFForceFieldHelpers (OTHER_FAULT) >> 55 - pyForceFieldHelpers (Failed) >> 56 - testDistGeomHelpers (OTHER_FAULT) >> 57 - pyDistGeom (Failed) >> 58 - testMolAlign (OTHER_FAULT) >> 59 - pyMolAlign (Failed)
Re: [Rdkit-discuss] RDKit build issues
Getting the right version of boost can be tricky. You can see our normal cmake incantation here as well as how we set RDBASE for tests https://github.com/rdkit/rdkit/blob/master/.travis.yml Note the -D Boost_NO_SYSTEM_PATHS=ON When running cmake, otherwise cmake can get very confused. Brian Kelley > On Apr 15, 2016, at 6:26 PM, Matthew Lardywrote: > > I'll add, that remembering that cmake and ccmake can produce different > outcomes I've gone back to trying cmake. But I can't overwrite the variables > in cmake. Here are the results from trying to specify them: > > Manually-specified variables were not used by the project: > > BOOST_DIR > BOOST_INCLUDE_DIR > BOOST_LIBRARY_DIR > BOOST_PYTHON_LIBRARY_DEBUG > BOOST_PYTHON_LIBRARY_RELEASE > BOOST_REGEX_LIBRARY_DEBUG > BOOST_REGEX_LIBRARY_RELEASE > BOOST_SERIALIZATION_LIBRARY_DEBUG > BOOST_SERIALIZATION_LIBRARY_RELEASE > BOOST_SYSTEM_LIBRARY_DEBUG > BOOST_SYSTEM_LIBRARY_RELEASE > BOOST_THREAD_LIBRARY_DEBUG > BOOST_THREAD_LIBRARY_RELEASE > > cmake, as per it's usual, picked up an ancient version of boost (which I want > to override). I can get around this with ccmake, but nothing that I compile > can pass all of the tests. If someone knows what these variables (shown in > ccmake) are called, I'd love to know. > > Thanks in advance! > Matthew > > >> On Fri, Apr 15, 2016 at 2:58 PM, Matthew Lardy wrote: >> Hi all, >> >> If someone has an insight I would love to hear it about how best to build >> RDKit from scratch. >> >> I am using the following to build the 2015.03 release: >> RedHat Linux version 6.4 >> gcc version 4.4.7 20120313 (Red Hat 4.4.7-11) (GCC) >> swig version 3.0.8 >> cmake version 3.0.0 >> boost version 1.53.0 (I've also tried boost 1.59.0) >> python version 2.7.7 >> >> I use ccmake and alter the things which need to be altered (as cmake always >> grabs the wrong things for me). My build compiles nicely, and without >> errors. But the tests are a mess. Here is the summary of my build: >> >> 26% tests passed, 87 tests failed out of 118 >> >> I've put the details of the tests, in the hope that someone sees a pattern >> that I do not, below. >> >> Before someone recommends a clean install that is not possible. I also do >> not have root, so I can't just wipe the machine and start over. I cannot >> use an RPM, so that is out too. Any ideas are welcome! >> >> Thanks, >> Matthew >> >> The following tests FAILED: >> 2 - testDataStructs (OTHER_FAULT) >> 3 - pyBV (Failed) >> 4 - pyDiscreteValueVect (Failed) >> 5 - pySparseIntVect (Failed) >> 7 - testGrid (OTHER_FAULT) >> 8 - testPyGeometry (Failed) >> 11 - pyAlignment (Failed) >> 14 - testMMFFForceField (OTHER_FAULT) >> 15 - pyForceFieldConstraints (Failed) >> 17 - pyDistGeom (Failed) >> 18 - graphmolTest1 (OTHER_FAULT) >> 21 - graphmolMolOpsTest (SEGFAULT) >> 23 - graphmoltestChirality (OTHER_FAULT) >> 24 - graphmoltestPickler (OTHER_FAULT) >> 26 - hanoiTest (OTHER_FAULT) >> 28 - testDepictor (OTHER_FAULT) >> 29 - pyDepictor (Failed) >> 32 - fileParsersTest1 (OTHER_FAULT) >> 33 - testMolSupplier (OTHER_FAULT) >> 34 - testMolWriter (OTHER_FAULT) >> 35 - testTplParser (OTHER_FAULT) >> 36 - testMol2ToMol (OTHER_FAULT) >> 38 - testReaction (OTHER_FAULT) >> 40 - pyChemReactions (Failed) >> 41 - testChemTransforms (OTHER_FAULT) >> 44 - testFragCatalog (OTHER_FAULT) >> 45 - pyFragCatalog (Failed) >> 46 - testDescriptors (OTHER_FAULT) >> 47 - pyMolDescriptors (Failed) >> 48 - testFingerprints (OTHER_FAULT) >> 50 - pyPartialCharges (Failed) >> 51 - testMolTransforms (OTHER_FAULT) >> 52 - pyMolTransforms (Failed) >> 53 - testMMFFForceFieldHelpers (OTHER_FAULT) >> 54 - testUFFForceFieldHelpers (OTHER_FAULT) >> 55 - pyForceFieldHelpers (Failed) >> 56 - testDistGeomHelpers (OTHER_FAULT) >> 57 - pyDistGeom (Failed) >> 58 - testMolAlign (OTHER_FAULT) >> 59 - pyMolAlign (Failed) >> 60 - testFeatures (OTHER_FAULT) >> 61 - pyChemicalFeatures (Failed) >> 62 - testShapeHelpers (OTHER_FAULT) >> 63 - pyShapeHelpers (Failed) >> 65 - pyMolCatalog (Failed) >> 66 - moldraw2DTest1 (OTHER_FAULT) >> 67 - pyMolDraw2D (Failed) >> 69 - pyFMCS (Failed) >> 72 - pyMolHash (Failed) >> 74 - pySLNParse (Failed) >> 75 - pyGraphMolWrap (Failed) >> 76 - pyTestConformerWrap (Failed) >> 79 - pyMatCalc (Failed) >> 80 - pySimDivPickers (Failed) >> 81 - pyRanker (Failed) >> 83 - pyFeatures (Failed) >>
Re: [Rdkit-discuss] RDKit build issues
This looks suspiciously like you haven't set RDBASE to point to the source directory when running the tests. The tests need to find the data directory with the test data. Brian Kelley > On Apr 15, 2016, at 6:26 PM, Matthew Lardywrote: > > I'll add, that remembering that cmake and ccmake can produce different > outcomes I've gone back to trying cmake. But I can't overwrite the variables > in cmake. Here are the results from trying to specify them: > > Manually-specified variables were not used by the project: > > BOOST_DIR > BOOST_INCLUDE_DIR > BOOST_LIBRARY_DIR > BOOST_PYTHON_LIBRARY_DEBUG > BOOST_PYTHON_LIBRARY_RELEASE > BOOST_REGEX_LIBRARY_DEBUG > BOOST_REGEX_LIBRARY_RELEASE > BOOST_SERIALIZATION_LIBRARY_DEBUG > BOOST_SERIALIZATION_LIBRARY_RELEASE > BOOST_SYSTEM_LIBRARY_DEBUG > BOOST_SYSTEM_LIBRARY_RELEASE > BOOST_THREAD_LIBRARY_DEBUG > BOOST_THREAD_LIBRARY_RELEASE > > cmake, as per it's usual, picked up an ancient version of boost (which I want > to override). I can get around this with ccmake, but nothing that I compile > can pass all of the tests. If someone knows what these variables (shown in > ccmake) are called, I'd love to know. > > Thanks in advance! > Matthew > > >> On Fri, Apr 15, 2016 at 2:58 PM, Matthew Lardy wrote: >> Hi all, >> >> If someone has an insight I would love to hear it about how best to build >> RDKit from scratch. >> >> I am using the following to build the 2015.03 release: >> RedHat Linux version 6.4 >> gcc version 4.4.7 20120313 (Red Hat 4.4.7-11) (GCC) >> swig version 3.0.8 >> cmake version 3.0.0 >> boost version 1.53.0 (I've also tried boost 1.59.0) >> python version 2.7.7 >> >> I use ccmake and alter the things which need to be altered (as cmake always >> grabs the wrong things for me). My build compiles nicely, and without >> errors. But the tests are a mess. Here is the summary of my build: >> >> 26% tests passed, 87 tests failed out of 118 >> >> I've put the details of the tests, in the hope that someone sees a pattern >> that I do not, below. >> >> Before someone recommends a clean install that is not possible. I also do >> not have root, so I can't just wipe the machine and start over. I cannot >> use an RPM, so that is out too. Any ideas are welcome! >> >> Thanks, >> Matthew >> >> The following tests FAILED: >> 2 - testDataStructs (OTHER_FAULT) >> 3 - pyBV (Failed) >> 4 - pyDiscreteValueVect (Failed) >> 5 - pySparseIntVect (Failed) >> 7 - testGrid (OTHER_FAULT) >> 8 - testPyGeometry (Failed) >> 11 - pyAlignment (Failed) >> 14 - testMMFFForceField (OTHER_FAULT) >> 15 - pyForceFieldConstraints (Failed) >> 17 - pyDistGeom (Failed) >> 18 - graphmolTest1 (OTHER_FAULT) >> 21 - graphmolMolOpsTest (SEGFAULT) >> 23 - graphmoltestChirality (OTHER_FAULT) >> 24 - graphmoltestPickler (OTHER_FAULT) >> 26 - hanoiTest (OTHER_FAULT) >> 28 - testDepictor (OTHER_FAULT) >> 29 - pyDepictor (Failed) >> 32 - fileParsersTest1 (OTHER_FAULT) >> 33 - testMolSupplier (OTHER_FAULT) >> 34 - testMolWriter (OTHER_FAULT) >> 35 - testTplParser (OTHER_FAULT) >> 36 - testMol2ToMol (OTHER_FAULT) >> 38 - testReaction (OTHER_FAULT) >> 40 - pyChemReactions (Failed) >> 41 - testChemTransforms (OTHER_FAULT) >> 44 - testFragCatalog (OTHER_FAULT) >> 45 - pyFragCatalog (Failed) >> 46 - testDescriptors (OTHER_FAULT) >> 47 - pyMolDescriptors (Failed) >> 48 - testFingerprints (OTHER_FAULT) >> 50 - pyPartialCharges (Failed) >> 51 - testMolTransforms (OTHER_FAULT) >> 52 - pyMolTransforms (Failed) >> 53 - testMMFFForceFieldHelpers (OTHER_FAULT) >> 54 - testUFFForceFieldHelpers (OTHER_FAULT) >> 55 - pyForceFieldHelpers (Failed) >> 56 - testDistGeomHelpers (OTHER_FAULT) >> 57 - pyDistGeom (Failed) >> 58 - testMolAlign (OTHER_FAULT) >> 59 - pyMolAlign (Failed) >> 60 - testFeatures (OTHER_FAULT) >> 61 - pyChemicalFeatures (Failed) >> 62 - testShapeHelpers (OTHER_FAULT) >> 63 - pyShapeHelpers (Failed) >> 65 - pyMolCatalog (Failed) >> 66 - moldraw2DTest1 (OTHER_FAULT) >> 67 - pyMolDraw2D (Failed) >> 69 - pyFMCS (Failed) >> 72 - pyMolHash (Failed) >> 74 - pySLNParse (Failed) >> 75 - pyGraphMolWrap (Failed) >> 76 - pyTestConformerWrap (Failed) >> 79 - pyMatCalc (Failed) >> 80 - pySimDivPickers (Failed) >> 81 - pyRanker (Failed) >> 83 - pyFeatures (Failed) >> 84 - JavaAromaticTests (Failed) >> 85 - JavaAtomPairsTests (Failed) >> 86 - JavaBasicMoleculeTests
Re: [Rdkit-discuss] RDKit build issues
I'll add, that remembering that cmake and ccmake can produce different outcomes I've gone back to trying cmake. But I can't overwrite the variables in cmake. Here are the results from trying to specify them: Manually-specified variables were not used by the project: BOOST_DIR BOOST_INCLUDE_DIR BOOST_LIBRARY_DIR BOOST_PYTHON_LIBRARY_DEBUG BOOST_PYTHON_LIBRARY_RELEASE BOOST_REGEX_LIBRARY_DEBUG BOOST_REGEX_LIBRARY_RELEASE BOOST_SERIALIZATION_LIBRARY_DEBUG BOOST_SERIALIZATION_LIBRARY_RELEASE BOOST_SYSTEM_LIBRARY_DEBUG BOOST_SYSTEM_LIBRARY_RELEASE BOOST_THREAD_LIBRARY_DEBUG BOOST_THREAD_LIBRARY_RELEASE cmake, as per it's usual, picked up an ancient version of boost (which I want to override). I can get around this with ccmake, but nothing that I compile can pass all of the tests. If someone knows what these variables (shown in ccmake) are called, I'd love to know. Thanks in advance! Matthew On Fri, Apr 15, 2016 at 2:58 PM, Matthew Lardywrote: > Hi all, > > If someone has an insight I would love to hear it about how best to build > RDKit from scratch. > > I am using the following to build the 2015.03 release: > RedHat Linux version 6.4 > gcc version 4.4.7 20120313 (Red Hat 4.4.7-11) (GCC) > swig version 3.0.8 > cmake version 3.0.0 > boost version 1.53.0 (I've also tried boost 1.59.0) > python version 2.7.7 > > I use ccmake and alter the things which need to be altered (as cmake > always grabs the wrong things for me). My build compiles nicely, and > without errors. But the tests are a mess. Here is the summary of my build: > > 26% tests passed, 87 tests failed out of 118 > > I've put the details of the tests, in the hope that someone sees a pattern > that I do not, below. > > Before someone recommends a clean install that is not possible. I also do > not have root, so I can't just wipe the machine and start over. I cannot > use an RPM, so that is out too. Any ideas are welcome! > > Thanks, > Matthew > > The following tests FAILED: > 2 - testDataStructs (OTHER_FAULT) > 3 - pyBV (Failed) > 4 - pyDiscreteValueVect (Failed) > 5 - pySparseIntVect (Failed) > 7 - testGrid (OTHER_FAULT) > 8 - testPyGeometry (Failed) > 11 - pyAlignment (Failed) > 14 - testMMFFForceField (OTHER_FAULT) > 15 - pyForceFieldConstraints (Failed) > 17 - pyDistGeom (Failed) > 18 - graphmolTest1 (OTHER_FAULT) > 21 - graphmolMolOpsTest (SEGFAULT) > 23 - graphmoltestChirality (OTHER_FAULT) > 24 - graphmoltestPickler (OTHER_FAULT) > 26 - hanoiTest (OTHER_FAULT) > 28 - testDepictor (OTHER_FAULT) > 29 - pyDepictor (Failed) > 32 - fileParsersTest1 (OTHER_FAULT) > 33 - testMolSupplier (OTHER_FAULT) > 34 - testMolWriter (OTHER_FAULT) > 35 - testTplParser (OTHER_FAULT) > 36 - testMol2ToMol (OTHER_FAULT) > 38 - testReaction (OTHER_FAULT) > 40 - pyChemReactions (Failed) > 41 - testChemTransforms (OTHER_FAULT) > 44 - testFragCatalog (OTHER_FAULT) > 45 - pyFragCatalog (Failed) > 46 - testDescriptors (OTHER_FAULT) > 47 - pyMolDescriptors (Failed) > 48 - testFingerprints (OTHER_FAULT) > 50 - pyPartialCharges (Failed) > 51 - testMolTransforms (OTHER_FAULT) > 52 - pyMolTransforms (Failed) > 53 - testMMFFForceFieldHelpers (OTHER_FAULT) > 54 - testUFFForceFieldHelpers (OTHER_FAULT) > 55 - pyForceFieldHelpers (Failed) > 56 - testDistGeomHelpers (OTHER_FAULT) > 57 - pyDistGeom (Failed) > 58 - testMolAlign (OTHER_FAULT) > 59 - pyMolAlign (Failed) > 60 - testFeatures (OTHER_FAULT) > 61 - pyChemicalFeatures (Failed) > 62 - testShapeHelpers (OTHER_FAULT) > 63 - pyShapeHelpers (Failed) > 65 - pyMolCatalog (Failed) > 66 - moldraw2DTest1 (OTHER_FAULT) > 67 - pyMolDraw2D (Failed) > 69 - pyFMCS (Failed) > 72 - pyMolHash (Failed) > 74 - pySLNParse (Failed) > 75 - pyGraphMolWrap (Failed) > 76 - pyTestConformerWrap (Failed) > 79 - pyMatCalc (Failed) > 80 - pySimDivPickers (Failed) > 81 - pyRanker (Failed) > 83 - pyFeatures (Failed) > 84 - JavaAromaticTests (Failed) > 85 - JavaAtomPairsTests (Failed) > 86 - JavaBasicMoleculeTests (Failed) > 87 - JavaBasicMolecule2Tests (Failed) > 88 - JavaChemAtomTests (Failed) > 89 - JavaChemBondTests (Failed) > 90 - JavaChemReactionTests (Failed) > 91 - JavaChemSmartsTests (Failed) > 92 - JavaChemTests (Failed) > 93 - JavaChemv2Tests (Failed) > 94 - JavaConformerTests (Failed) > 95 - JavaDescriptorTests (Failed) > 96 -