You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
The GLM CMake was not modernised during the big CMake modernisation, so it still relies on global cmake variables rather than using a header only target.
This should be improved to match other header only targets.
Unfortunately, recent GLM seems to have dropped the use of version numbers / hasn't made a new versioned relese since april 2020, with almost 200 commits since this release, so using CMake to check the correct GLM was found might not be viable.
In addition, the vis repo GLM should be modenised together with this. Ideally with both repos requireing the same glm version, and only downloading it once.
Alonside the use of targets, the main flamegpu library must also inform RTC of include paths. This is currently handled by a macro definiation GLM_PATH, forwarding the include path from CMake. This is non portable, so we need to address this (part of #260, maybe #607).
Further more, 2.0.0a1 USE_GLM builds only work when visualiseation is enabled. This will be fixed quickly in the short term with a very minor change, to be improved at a later date.
Modernise GLM use, to use an imported interface target (see FindJitify.cmake)
Ideally find existing GLM versions and check if it is sufficiently new. This may not be possible as GLM seems to have stopped making new versioned releases
Update vis repo GLM to match
Use the same GLM in both repositories if required. I.e. only download if GLM was not already in _deps. Without usable version numbers it will be hard to support this and allow use-provided glm
Optionally: Expand CI to include a USE_GLM=ON build(s) to ensure it works in the future.
The text was updated successfully, but these errors were encountered:
The GLM CMake was not modernised during the big CMake modernisation, so it still relies on global cmake variables rather than using a header only target.
This should be improved to match other header only targets.
Unfortunately, recent GLM seems to have dropped the use of version numbers / hasn't made a new versioned relese since april 2020, with almost 200 commits since this release, so using CMake to check the correct GLM was found might not be viable.
In addition, the vis repo GLM should be modenised together with this. Ideally with both repos requireing the same glm version, and only downloading it once.
Alonside the use of targets, the main flamegpu library must also inform RTC of include paths. This is currently handled by a macro definiation
GLM_PATH
, forwarding the include path from CMake. This is non portable, so we need to address this (part of #260, maybe #607).Further more, 2.0.0a1
USE_GLM
builds only work when visualiseation is enabled. This will be fixed quickly in the short term with a very minor change, to be improved at a later date.imported interface
target (seeFindJitify.cmake
)The text was updated successfully, but these errors were encountered: