Re: [Rdkit-discuss] Compiling on MacOSX
Dear all, I've added a page to the wiki with build instructions for MacOS/X: http://code.google.com/p/rdkit/wiki/BuildingOnMacOs Andrew, these include your troubleshooting information from below. Please let me know if you want me to edit, rewrite, or remove your name. :-) -greg On Sat, Nov 22, 2008 at 5:11 AM, Andrew Dalke da...@dalkescientific.com wrote: On Nov 21, 2008, at 9:20 AM, Greg Landrum wrote: The Q3 release (and current svn version, of course) do build (and work!) on MacOS. I've compiled it on my 10.4 machine but it required some changes that Greg put into svn and which aren't in the Q3 release. We had that discussion by private email, but it should probably have been done on- list. There are still two things which need to be done to compile from svn on my box: 1) I edited Jamroot to point to my numpy include directory. was: toolsetdarwin:include/System/Library/Frameworks/ Python.framework/Versions/2.5/Extras/lib/python/numpy/core/include changed to: toolsetdarwin:include/Library/Frameworks/Python.framework/ Versions/2.5/lib/python2.5/site-packages/numpy/core/include 2) 10.4 uses an old version of yacc (v1.28 from 1999!), which for some reason Boost doesn't know how to handle. Greg's looking in to how to fix this automatically, but another solution might be to assume that the number of people compiling with 10.4 is dwindling and not worth worrying about. The problem occurs when converting the yacc grammars for SMILES and SMARTS. Here's a diagnostic error message: GraphMol/SmilesParse/smiles.ll:30:26: error: smiles.tab.hpp: No such file or directory GraphMol/SmilesParse/smiles.ll: In function 'int yysmiles_lex()': GraphMol/SmilesParse/smiles.ll:44: error: 'CHI_CLASS_TOKEN' was not declared in this scope GraphMol/SmilesParse/smiles.ll:46: error: 'AT_TOKEN' was not declared in this scope The problem is the created output file is smiles.tab.cpp.h instead of smiles.tab.hpp [josiah:~/ftps/RDKit_Q32008_1/Code] dalke% find . -name 'smiles.tab*' ./GraphMol/SmilesParse/bin/darwin/release/macosx-version-10.4/ threading-multi/smiles.tab.cpp ./GraphMol/SmilesParse/bin/darwin/release/macosx-version-10.4/ threading-multi/smiles.tab.cpp.h ./GraphMol/SmilesParse/bin/darwin/release/macosx-version-10.4/ threading-multi/smiles.tab.o so it looks like the output was .cpp.h instead of .hpp. I manually copied it to the .hpp form and it seems to have worked - I ended up with problems with smarts.tab.hpp instead: That is, I renamed the file, recompiled, and ended up with similar problems with the SMARTS grammar. I then renamed smarts.tab.cpp.h to smarts.tab.hpp and recompiled. That gets me to completion. Andrew da...@dalkescientific.com - This SF.Net email is sponsored by the Moblin Your Move Developer's challenge Build the coolest Linux based applications with Moblin SDK win great prizes Grand prize is a trip for two to an Open Source event anywhere in the world http://moblin-contest.org/redirect.php?banner_id=100url=/ ___ Rdkit-discuss mailing list Rdkit-discuss@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/rdkit-discuss
Re: [Rdkit-discuss] RDKit - DbCLI
Dear Kirk, On Fri, Nov 21, 2008 at 12:38 AM, Robert DeLisle rkdeli...@gmail.com wrote: After running through the process with exception handling in place I was able to isolate 10 structures that were being problematic. All of them had at least one bond designated as 0 order in the SD file - much as you found for some of the other structures previously. I assume that these passed the initial import step but are failing upon descriptor generation for obvious reasons. I suppose the only request that I have is for more graceful error handling. I've attached my (admittedly sloppy) version of CreateDB.py showing what I did to isolate the errors. The problem here was in the mol file parser: it was not correctly setting up bonds that have order 0. Now it generates a warning (order 0 isn't technically allowed by the ctab spec) and sets the bond up correctly. I also added some error checking to handle other bogus bond orders. This was entered as issue 2337369 (https://sourceforge.net/tracker2/?func=detailaid=2337369group_id=160139atid=814650) and fixed in rev892. -greg
Re: [Rdkit-discuss] Compiling on MacOSX
Greg: Please let me know if you want me to edit, rewrite, or remove your name. :-) No worries; not a problem. But why is there so much green and blue text? ... Ahh, WikiMarkup. Andrew da...@dalkescientific.com
Re: [Rdkit-discuss] Compiling on MacOSX
On Mon, Nov 24, 2008 at 3:05 PM, Andrew Dalke da...@dalkescientific.com wrote: Greg: Please let me know if you want me to edit, rewrite, or remove your name. :-) No worries; not a problem. But why is there so much green and blue text? ... Ahh, WikiMarkup. yeah, the google wiki formatter thinks that's a code sample is trying (and failing, miserably) to highlight it. -greg
Re: [Rdkit-discuss] RDKit - DbCLI
Fantastic! Thanks, Greg! After I got things working I've been able to generate a database and do some preliminary searches. I'm impressed at how quickly I can search ~100,000 compounds with SMARTS patterns. I have a feeling this one is going to get a lot of use. -Kirk On Mon, Nov 24, 2008 at 12:45 AM, Greg Landrum greg.land...@gmail.comwrote: Dear Kirk, On Fri, Nov 21, 2008 at 12:38 AM, Robert DeLisle rkdeli...@gmail.com wrote: After running through the process with exception handling in place I was able to isolate 10 structures that were being problematic. All of them had at least one bond designated as 0 order in the SD file - much as you found for some of the other structures previously. I assume that these passed the initial import step but are failing upon descriptor generation for obvious reasons. I suppose the only request that I have is for more graceful error handling. I've attached my (admittedly sloppy) version of CreateDB.py showing what I did to isolate the errors. The problem here was in the mol file parser: it was not correctly setting up bonds that have order 0. Now it generates a warning (order 0 isn't technically allowed by the ctab spec) and sets the bond up correctly. I also added some error checking to handle other bogus bond orders. This was entered as issue 2337369 ( https://sourceforge.net/tracker2/?func=detailaid=2337369group_id=160139atid=814650 ) and fixed in rev892. -greg
Re: [Rdkit-discuss] RDKit - DbCLI
On Mon, Nov 24, 2008 at 5:11 PM, Robert DeLisle rkdeli...@gmail.com wrote: Fantastic! Thanks, Greg! After I got things working I've been able to generate a database and do some preliminary searches. I'm impressed at how quickly I can search ~100,000 compounds with SMARTS patterns. I have a feeling this one is going to get a lot of use. glad to hear it looks useful. The search speed isn't terrible as it is, but it could be made a ton faster by using a substructure fingerprint. But that's for a later version. -greg
