Closed varunagrawal closed 2 years ago
Let’s stick to gtdynamics. Two questions: 1) how do you know it works? Add evidence in PR comment. 2) add some statement that makes me believe it does not affect anyone negatively Happy to approve after that - let me know by e-mail.
This PR adds support for making
gtdynamics
a cmake package so we can dofind_package(gtdynamics)
for downstream applications. Fixes #5.I need this for my own LRSE project and it's been long overdue. To include gtdynamics, all I needed to do was
find_package(gtdynamics REQUIRED)
and when I print the version as${gtdynamics_VERSION}
, I get the following output:This is a CMake update and no C++ changes are made whatsoever. Moreover, it includes commands for configuring and installing mainly 3 files:
gtdynamics-exports.cmake
,gtdynamicsConfig.cmake
andgtdynamicsConfigVersion.cmake
. Thus, there should be no breakage in the normal process of installation from source. Even the CI should be evidence of this since we install from source.Question: Should we stick to
gtdynamics
or convert the package name toGTDynamics
? The former means we have CMake variables of the formgtdynamics_VERSION
which is fine IMO, but we can doGTDynamics_VERSION
instead if you'd like. GTSAM doesGTSAM_VERSION
but it is also a shorter name.