-
Notifications
You must be signed in to change notification settings - Fork 52
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
Hardening internal fmt/camp TPL install config cases #870
Conversation
@@ -147,6 +147,7 @@ if (NOT TARGET fmt::fmt) | |||
if (NOT EXISTS ${PROJECT_SOURCE_DIR}/src/tpl/umpire/fmt/CMakeLists.txt) | |||
message(FATAL_ERROR "fmt submodule not initialized. Run 'git submodule update --init --recursive' in the git repository or set fmt_DIR to use an external build of fmt.") | |||
else () | |||
set(fmt_DIR ${CMAKE_INSTALL_PREFIX} CACHE PATH "") |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
If I understand correctly, fmt_DIR is used here: https://github.com/LLNL/Umpire/blob/develop/umpire-config.cmake.in#L37
When using an external fmt the call to find_package will set fmt_DIR for us, which is searched. However, I'm confused why this doesn't work since PACKAGE_CMAKE_INSTALL_PREFIX should be the same as CMAKE_INSTALL_PREFIX: https://github.com/LLNL/Umpire/blob/develop/umpire-config.cmake.in#L45
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
hmm... that is an excellent point. Also explains why I couldn't reproduce the original issue. I will need to see if I can force this to fail..
man @adayton1 why must you rain on my parade.
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
☔ 🌧️
I tried to test this branch in my code, but I also need the fixes in |
We could try merging this branch into the blt-tpl-exports branch and then you could test... |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
LGTM
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
LGTM
The nvcc Docker test is failing because "no space left on device".. how can I resolve that? |
@kab163 we've had similar issues with RAJA. The cause is that the containers contain a bunch of Spack stuff that is not needed and azure reduced the amount of disk space allowed for free accounts. Chris was working on building a new container repo to address this but I don't know what the status of that is. For RAJA, we turned off nvcc and rocm checks in azure since we are covering those cases in LC GItLab CI. |
There is an issue when using an internal build of Umpire that the fmt_DIR is not set correctly and users have to point cmake at where Umpire has it's fmt install. To fix that, we have to set the fmt_DIR in the tpl CMakeLists file