[AMBER-Developers] Amber and gcc6

From: David A Case <david.case.rutgers.edu>
Date: Mon, 23 May 2016 17:57:29 -0400

Hi everyone:

Gcc version 6 was released at the end of April, so we can expect people to
start using it on our code soon. Here's what I know so far:

1. reduce and pytraj are the two parts that don't build for me (using
the MacPorts version of gcc 6.1). In reduce, if I add "#define OLD_STD_HDRS 1"
to pdb++.h, that gets rid of many (but not all) the problems. Not sure that
this is the best fix; these changes mostly arise from changes in how
iostreams are handled, particular in casting iostream results to things like
bool variables. Since iostreams are evil anyway, and I am not much of a C++
person, others will have to look at this.

2. In line 12 of cphstats/utilities.cpp, changing "return f" to
"return static_cast<bool>( f );" seems to help. But Dan says he found a
better way to handle the same problem in cpptraj, so there could be another
solution there.

3. If one skips reducen and pytraj, everything builds.

4. Serial test results: good, except for the very common pytraj problem:
dlopen(/home/case/amber16/lib/python2.7/site-packages/pytraj/topology.so, 2):
Symbol not found: __ZN8Topology7AddBondEii

Plus some file diffs in leap/4PY5 and leap/BSC1 that I get with almost
all compilers: has anyone else seen these?

Anyone wishing to tackle the pytraj and reduce problems are welcome to have a


AMBER-Developers mailing list
Received on Mon May 23 2016 - 15:00:04 PDT
Custom Search