Quoting Josh Andler <scislac@...400...>:
If it's already in use in 2geom and liked, I'd vote for it just because of that alone.
Cheers, Josh
Yes, Google Test and Boost Test are close enough that the use by 2geom might be a strong deciding factor.
To summarize some of the differences:
Google Test: * does not use exceptions. Among other things this means that if a function is called that might result in a test failing, the calling test will need to add an extra line to check status afterwards. * intended to be copied into the using project's source tree * needs the additional 'Google Mock' library for certain things such as comparing collections
Boost Test: * uses exceptions to terminate tests * use is with pre-built libs/source that are not copied in-tree (which is an external dependency on that Boost library) * does not have an academic/artificial distinction between 'Mock' objects and general fakes/stubs
So as long as it looks good to switch to Google Test, I'd suggest copying in Google Mock to our source, since it is a superset and contains a copy of the full Google Test sources. We should keep the actual gmock use to a minimum (even Google themselves warn against Mock overuse), but the collections support alone is extremely useful.