Reject find_package calls with non-standard capitalization

[mattjala]

HDF5 installs hdf5-config.cmake, and CMake's config-mode search accepts either Config.cmake or -config.cmake, so find_package() locates HDF5 for any capitalization the caller provides (HDF5, hdf5, HdF5). CMake then records the requested components under the exact capitalization that was used, in _FIND_COMPONENTS.

The config reads that state as ${HDF5_PACKAGE_NAME}_FIND_COMPONENTS, and HDF5_PACKAGE_NAME is the uppercased package name, so it always reads the specific variable HDF5_FIND_COMPONENTS. Any capitalization other than HDF5 has its component list read back as empty. The package is still found, but the config concludes that no components were requested and applies its defaults.

This PR swaps ${HDF5_PACKAGE_NAME} for ${CMAKE_FIND_PACKAGE_NAME} in the names of variables used by CMake's mechanisms in order to correctly retain the components list for any provided capitalization pattern for HDF5.

EDIT: The problem this PR addressed is now resolved by rejecting incorrect capitalizations instead of doing work to make sure they still get the correct configuration information.

[github-actions]

Review Checklist

This PR touches the following areas. Each needs a sign-off
from its listed owners before merging.

• config
  • @jhendersonHDF (manually added) — approval required

[Copilot]

Pull request overview

This PR updates the installed hdf5-config.cmake template to correctly respect find_package() component requests regardless of how the caller capitalizes the package name (e.g., HDF5, hdf5, HdF5), by using CMAKE_FIND_PACKAGE_NAME for CMake’s *_FIND_COMPONENTS-related variables.

Changes:

• Switch component handling from ${HDF5_PACKAGE_NAME}_FIND_COMPONENTS to ${CMAKE_FIND_PACKAGE_NAME}_FIND_COMPONENTS so the caller’s requested component list is preserved for any capitalization.
• Update *_NOT_FOUND_MESSAGE assignment to use ${CMAKE_FIND_PACKAGE_NAME} so error reporting aligns with the caller’s package name capitalization.

Suppressed comments (1)

config/install/hdf5-config.cmake.in:228

• When libaec targets can’t be loaded, ${HDF5_PACKAGE_NAME}_FOUND is set to FALSE, but for config-mode packages CMake checks ${CMAKE_FIND_PACKAGE_NAME}_FOUND (the capitalization used in find_package()). This can cause find_package(hdf5 REQUIRED) to succeed even though the config returns early.

      set (${HDF5_PACKAGE_NAME}_FOUND FALSE)
      set (${CMAKE_FIND_PACKAGE_NAME}_NOT_FOUND_MESSAGE "Couldn't locate libaec targets file ${_libaec_targets_file_name}")

---

💡 Add a code-review agent skill or configure MCP servers for context-aware, tailored reviews. Learn more in the docs.

[jhendersonHDF]

I may need to think about this more, but this is a case where I might actually suggest simply rejecting any package name that isn't exactly HDF5 in our configuration file instead of trying to work around the problem. Both find_package (hdf5 MODULE REQUIRED) and find_package (HdF5 MODULE REQUIRED) will fail because the module in CMake is FindHDF5.cmake and the convention for variables set by that is HDF5_XXX, so we typically would want to always match that convention in our configuration file for variables set for compatibility. Something like hdf5_XXX would only confuse the issue.