I think it is time to revisit this topic. We will shortly upgrade SWIG to 3.0.2. The main motivation for that upgrade is C++11 support, and we ought to able to take advantage of it.
The standard described on this page endorses the use of "long long". In cases where the actual bit width of a variable is an issue, do we prefer the use of the <cstdint> types? E.g., the use of "int64_t" in preference to "long long"?
4 Comments
Russell Owen
I think it is time to revisit this topic. We will shortly upgrade SWIG to 3.0.2. The main motivation for that upgrade is C++11 support, and we ought to able to take advantage of it.
Gregory Dubois-Felsmann
The standard described on this page endorses the use of "
long long
". In cases where the actual bit width of a variable is an issue, do we prefer the use of the<cstdint>
types? E.g., the use of "int64_t
" in preference to "long long
"?Gregory Dubois-Felsmann
See for instance http://en.cppreference.com/w/cpp/header/cstdint
Kian-Tat Lim
Unfortunately,
cstdint
is not supported by RHEL6 as discovered by Serge in https://github.com/lsst/sphgeom/pull/1Alternatives seem to be
stdint.h
orboost/cstdint.hpp
.