Sherm,
Yes, that's what I was leading to at least: optionally neither build nor test static libraries.
Use case would be a user who gets the source and is not modifying it, and (s)he knows which variation he needs (e.g. OpenSim using dynamic libs only).
I understand the main concern is that we need to make sure we can link executables, but that's more of a development time issue on our side.
I don't feel so strongly about it, however, since this scenario may not continue for the long term.
-Ayman
Do we need all those "Static" test cases?
- Ayman Habib
- Posts: 2248
- Joined: Fri Apr 01, 2005 12:24 pm
- Michael Sherman
- Posts: 807
- Joined: Fri Apr 01, 2005 6:05 pm
RE: Do we need all those "Static" test cases?
To close out this thread, the answer appears to be "yes, we do need all those static test cases!". However, I have now modified the CMake scripts so that you can turn off static, dynamic, or all tests selectively.
The nightly builds will continue building both static and dynamic versions of everything, because there are still a significant number of failures revealed by one kind but not the other.
The nightly builds will continue building both static and dynamic versions of everything, because there are still a significant number of failures revealed by one kind but not the other.