On 2/9/2013 12:35 PM, William G. Scott wrote:
> On Feb 9, 2013, at 9:16 AM, Martin Costabel <costabel@...>
>> On 9/02/13 17:22, William G. Scott wrote:
>>> Hi folks (Martin et al.):
>>> I'm trying to put together a fink package for rdkit, which is a
>>> new dependency for coot (which I maintain). rdkit requires a
>>> fairly recent version of boost. I can build it using version
>>> 1.53. Fink has 1.35, which is a bit long in the tooth. (I also
>>> tried using boost1.46.1.cmake. I got nowhere, but this might be
>>> simply the vastness of my own ignorance.)
>>> In any case, would it be possible to get a boost1.53 package ?
>>> Right now I am building the required library within my nascent
>>> rdkit package, but this is a bandaid approach.
>> just yesterday, someone sent me a package description for 1.52. As
>> usual, there are problems with the boost-python libs and the build
>> system has slightly changed. But maybe this might be useful for
>> 1.53, too.
>> Since I haven't looked at this for a very long time (boost-cmake
>> seemed to be the wave of the future), I cannot tell how long it
>> will take me to get something acceptable, but I'll have a look. The
>> users of the boost libraries seem to fall into one of two classes:
>> Those who don't care about python and those who *only* want the
>> libboost-python libs and don't care about the rest. Maybe two
>> separate package descriptions might simplify the update process.
The main problem with boost.cmake is that it links to system-python
Framework, and if you have something that needs boost and a pymod, then
things don't work all that well.
With that in mind, having 36+ inches of snow on the ground gave me some
free time, and I updated a boost package I've been sort of holding in
the background to 1.53. It uses the same package layout as the 1.35
version. The biggest difference in the build is that boost-jam is no
longer used because of some version mismatch, so it has to bootstrap
first, and then use an internal (b2) build tool, plus it has a bunch of
You can see my work here:
I've only tested it on 10.7/Xcode4.6, and I don't know how it will
behave on other system, especially since there are some games played
with the compiler (which might not be necessary).