Hi Greg,

I appreciate your taking the time to look at this.

I have g++ 4.3, so that might be the problem. However, I downloaded
dynamic_bitset.hpp from SVN, and I used it to replace the 3 instances of the
file in my $BOOSTHOME. However, this did not solve the problem.

Many are still failing, such as

...failed gcc.compile.c++
> GraphMol/Wrap/bin/gcc-4.3.3/release/address-model-64/threading-multi/MolOps.o...
> ...skipped
> <pGraphMol/Wrap/bin/gcc-4.3.3/release/address-model-64/threading-multi>rdmolops.so
> for lack of
> <pGraphMol/Wrap/bin/gcc-4.3.3/release/address-model-64/threading-multi>MolOps.o...
> ...skipped
> <pGraphMol/Wrap/bin/gcc-4.3.3/release/address-model-64/threading-multi>rdmolfiles.so
> for lack of
> <pGraphMol/FileParsers/bin/gcc-4.3.3/release/address-model-64/threading-multi>libFileParsers.so...
> ...skipped <p/usr/local/src/RDKit_Q22009_1/rdkit/Chem>rdchem.so for lack of
> <pGraphMol/Wrap/bin/gcc-4.3.3/release/address-model-64/threading-multi>rdchem.so...
> ...skipped <p/usr/local/src/RDKit_Q22009_1/rdkit/Chem>rdmolops.so for lack
> of
> <pGraphMol/Wrap/bin/gcc-4.3.3/release/address-model-64/threading-multi>rdmolops.so...
> ...skipped <p/usr/local/src/RDKit_Q22009_1/rdkit/Chem>rdmolfiles.so for
> lack of
> <pGraphMol/Wrap/bin/gcc-4.3.3/release/address-model-64/threading-multi>rdmolfiles.so...
> gcc.compile.c++
> GraphMol/FragCatalog/Wrap/bin/gcc-4.3.3/release/address-model-64/threading-multi/FragCatParams.o
> In file included from ./DataStructs/ExplicitBitVect.h:9,
>                  from ./DataStructs/BitVects.h:15,
>                  from GraphMol/FragCatalog/Wrap/FragCatParams.cpp:10:
>

Any further suggestions?

-Ari



On Mon, Jul 6, 2009 at 4:21 PM, Greg Landrum <[email protected]> wrote:

> On Mon, Jul 6, 2009 at 9:28 PM, Aryeh
> Gold-Parker<[email protected]> wrote:
> > I am building from the source distribution. I have included a much longer
> > description of one of the many errors.
> >
>
> [snip]
>
> Most of these don't show enough detail for me to figure out exactly
> what happened, but this one does:
>
> >> DataManip/MetricMatrixCalc/bin/gcc-4.3.3/release/threading-multi
> >> gcc.compile.c++
> >>
> DataManip/MetricMatrixCalc/bin/gcc-4.3.3/release/threading-multi/testMatCalc.o
> >> In file included from
> >> /usr/local/src/boost_1_39_0/boost/dynamic_bitset.hpp:15,
> >>                  from ./DataStructs/ExplicitBitVect.h:9,
> >>                  from ./DataStructs/BitVects.h:15,
> >>                  from ./DataStructs/BitOps.h:19,
> >>                  from DataManip/MetricMatrixCalc/MetricFuncs.h:9,
> >>                  from DataManip/MetricMatrixCalc/testMatCalc.cpp:7:
> >> /usr/local/src/boost_1_39_0/boost/dynamic_bitset/dynamic_bitset.hpp: In
> >> member function ‘size_t boost::dynamic_bitset<Block, Allocator>::count()
> >> const’:
> >>
> /usr/local/src/boost_1_39_0/boost/dynamic_bitset/dynamic_bitset.hpp:1015:
> >> error: ‘mode’ cannot appear in a constant-expression
> >>
> /usr/local/src/boost_1_39_0/boost/dynamic_bitset/dynamic_bitset.hpp:1015:
> >> error: template argument 1 is invalid
> >>
> /usr/local/src/boost_1_39_0/boost/dynamic_bitset/dynamic_bitset.hpp:1015:
> >> error: expected `>' before ‘*’ token
> >>
> /usr/local/src/boost_1_39_0/boost/dynamic_bitset/dynamic_bitset.hpp:1015:
> >> error: expected `(' before ‘*’ token
> >>
> /usr/local/src/boost_1_39_0/boost/dynamic_bitset/dynamic_bitset.hpp:1015:
> >> error: expected primary-expression before ‘>’ token
>
> This looks like a known bug in g++ that has theoretically been fixed:
> https://bugs.launchpad.net/ubuntu/+source/gcc-4.3/+bug/354305
> Not sure if/when your packager is going to distribute those changes.
>
> As the thread above indicates, there's also a fix checked in for
> boost, but it's not in the 1.39 release. If you can't find a fixed
> version of g++, you might try downloading the dynamic_bitset header
> from boost svn:
>
> https://svn.boost.org/trac/boost/export/53050/trunk/boost/dynamic_bitset/dynamic_bitset.hpp
>
> > Tell me what else might be of use. I can run it again and have it log
> into a
> > text file if that would be helpful.
>
> I believe this should be enough.
>
> -greg
>

Reply via email to