[Boost-users] Why we are still using boost 1.37

...

But I'm curious to know why you are still using 1.42.

Well, I'm still using 1.37 for a large astronomical software project (the LSST). To quote one of the developers:

There have been changes to boost::serialization between these two versions that break our persistence code. The issues appear in test segfaults for packages at the afw level and higher.

We could presumably find the problem in our code (or what's changed in boost::serialization) but that's real work. It's easier to patch boost::gil 1.37 to work with gcc 4.4 (or was it 4.5)? R