Add Windows SHLWAPI lib to public interface #4701
Conversation
byrnHDF
added
Merge - To 1.14
Priority - 0. Blocker ⛔
byrnHDF
requested review from
lrknox,
derobins,
fortnern,
jhendersonHDF,
qkoziol,
vchoi-hdfgroup,
bmribler,
glennsong09,
mattjala and
brtnfld
as code owners
|
I'd definitely still like to know if we're doing something wrong since I don't understand the reason why this needs to be in the public link interface. I'm guessing that it has to do with functions which are meant to be library-private not having restricted visibility so they're in the link interface for anything which links to HDF5. |
see https://stackoverflow.com/questions/26037954/cmake-target-link-libraries-interface-dependencies |
That just seems to reinforce the idea that this shouldn't be a public link because there's no reason the functions that use the SHLWAPI lib should be in the public link interface. The library was linked against due to use of StrStrIA, but this is used in a function that isn't meant to be used publicly. It's in H5private.h through H5system.c and is meant for internal library use. But since we don't restrict the visibility of library-wide private functions, I'm guessing it ends up in the link interface unintentionally. |
|
If we want a definitive answer, we should ask on CMake's discourse https://discourse.cmake.org/ |
|
I could make it an INTERFACE instead of PUBLIC. |
We need to link against the shlwapi lib anyway, so I don't think INTERFACE is correct since "Libraries following INTERFACE are appended to the link interface and are not used for linking <target>." |
|
This will likely have to remain a PUBLIC link for now since it would take some time to figure out why shlwapi is a transitive dependency currently and some work to resolve. However, I'm not convinced that PUBLIC is ultimately correct and can't see how shlwapi is anything but a private dependency. |
You can see the GitHub issue linked on the right side under the "Development" section. Here's the list of GitHub keywords in a description that will automatically link an issue: https://docs.github.com/en/issues/tracking-your-work-with-issues/linking-a-pull-request-to-an-issue#linking-a-pull-request-to-an-issue-using-a-keyword |
|
INTERFACE didn't work - reverting back to PUBLIC |
src/CMakeLists.txt
Outdated
| @@ -1099,7 +1099,7 @@ if (BUILD_STATIC_LIBS) | |||
| TARGET_C_PROPERTIES (${HDF5_LIB_TARGET} STATIC) | |||
| target_link_libraries (${HDF5_LIB_TARGET} | |||
| PRIVATE ${LINK_LIBS} ${LINK_COMP_LIBS} | |||
| PUBLIC "$<$<NOT:$<PLATFORM_ID:Windows>>:${CMAKE_DL_LIBS}>" "$<$<BOOL:${HDF5_ENABLE_PARALLEL}>:MPI::MPI_C>" | |||
| PUBLIC "$<$<PLATFORM_ID:Windows>:${SHLWAPI_LIB}>" "$<$<NOT:$<PLATFORM_ID:Windows>>:${CMAKE_DL_LIBS}>" "$<$<BOOL:${HDF5_ENABLE_PARALLEL}>:MPI::MPI_C>" | |||
There was a problem hiding this comment.
Can we have a LINK_LIBS-style variable for public libs, just so we don't have to introduce unnecessary extra variables like SHLWAPI_LIB above?
There was a problem hiding this comment.
Sounds like a good idea
There was a problem hiding this comment.
Approving, though I hope two things come after at some point in the future:
- We should figure out why this dependency had to be made public since the library is really only used by library-internal functions and not exposed publicly in any way (at least not intentionally)
- Would be nice to have a public
LINK_LIBSstyle variable for public libraries, but that should probably come in a separate PR
|
* Warning fix (#4682) * warning fix * warning fix * CMake link line needs to use new HDF5_ENABLE_THREADS (#4685) * Correct the properties for using THREADS library (#4690) * Bump the github-actions group with 5 updates (#4688) Bumps the github-actions group with 5 updates: | Package | From | To | | --- | --- | --- | | [actions/download-artifact](https://github.com/actions/download-artifact) | `4.1.7` | `4.1.8` | | [DoozyX/clang-format-lint-action](https://github.com/doozyx/clang-format-lint-action) | `0.13` | `0.17` | | [softprops/action-gh-release](https://github.com/softprops/action-gh-release) | `2.0.6` | `2.0.8` | | [ossf/scorecard-action](https://github.com/ossf/scorecard-action) | `2.3.3` | `2.4.0` | | [github/codeql-action](https://github.com/github/codeql-action) | `3.25.11` | `3.25.15` | Updates `actions/download-artifact` from 4.1.7 to 4.1.8 - [Release notes](https://github.com/actions/download-artifact/releases) - [Commits](actions/download-artifact@65a9edc...fa0a91b) Updates `DoozyX/clang-format-lint-action` from 0.13 to 0.17 - [Release notes](https://github.com/doozyx/clang-format-lint-action/releases) - [Commits](DoozyX/clang-format-lint-action@v0.13...v0.17) Updates `softprops/action-gh-release` from 2.0.6 to 2.0.8 - [Release notes](https://github.com/softprops/action-gh-release/releases) - [Changelog](https://github.com/softprops/action-gh-release/blob/master/CHANGELOG.md) - [Commits](softprops/action-gh-release@a74c6b7...c062e08) Updates `ossf/scorecard-action` from 2.3.3 to 2.4.0 - [Release notes](https://github.com/ossf/scorecard-action/releases) - [Changelog](https://github.com/ossf/scorecard-action/blob/main/RELEASE.md) - [Commits](ossf/scorecard-action@dc50aa9...62b2cac) Updates `github/codeql-action` from 3.25.11 to 3.25.15 - [Release notes](https://github.com/github/codeql-action/releases) - [Changelog](https://github.com/github/codeql-action/blob/main/CHANGELOG.md) - [Commits](github/codeql-action@b611370...afb54ba) --- updated-dependencies: - dependency-name: actions/download-artifact dependency-type: direct:production update-type: version-update:semver-patch dependency-group: github-actions - dependency-name: DoozyX/clang-format-lint-action dependency-type: direct:production update-type: version-update:semver-minor dependency-group: github-actions - dependency-name: softprops/action-gh-release dependency-type: direct:production update-type: version-update:semver-patch dependency-group: github-actions - dependency-name: ossf/scorecard-action dependency-type: direct:production update-type: version-update:semver-minor dependency-group: github-actions - dependency-name: github/codeql-action dependency-type: direct:production update-type: version-update:semver-patch dependency-group: github-actions ... * Fix segfault when closing datatype during failure in H5Topen2 (#4683) * Rework Dynamic Analysis and sanitize testing (#4681) * Ignore predetermined failing test and check pointer before use * Rework Analysis process * Remove another H5E_BEGIN/END_TRY within the library (#4675) * Update logic for (deprecated) H5Gget_objinfo() call to eliminate H5E_BEGIN_TRY * Handle case for '.' at the end of a path * Drop H5E_BEGIN/END_TRY and just check the error return from H5I_clear_types() (#4694) Original case that the change in commit 2dc738a no longer applies. * Add check of returned value from API calls. (#4702) These were found while investigating GH-4672, but they were not related to GH-4672. * Add mac dmg binary and remove old macos-13 workflows (#4699) * Add Windows SHLWAPI lib to public interface (#4701) * Use local variable in btree2 and print value (#4679) * Correct logic * Technically, level 1 Express could skip tests * Add windows signing (#4703) * Add tests for H5R get name APIs (#4657) Added functionality tests for the following APIs: H5Rget_file_name H5Rget_obj_name H5Rget_attr_name Also removed "+1" when returning a name length in H5R__get_attr_name(). The exter "+1" gave an incorrect value for the length of the referenced object's attribute name. Fixed GH-4447 * Fix Fortran test The C API H5Rget_attr_name incorrectly added 1 to the length of the referenced object's attribute name, so the Fortran API h5rget_attr_name_f removed 1 from the returned value to accommodate the incorrectness. This PR fixes H5Rget_attr_name so this workaround in h5rget_attr_name_f is no longer needed. * Add test H5Aget_name against H5Rget_attr_name * Replace Visual Studio ???? with 2022 in MSI README file (#4709) * Change logic for checking secrets exists (#4711) * Change osx refs to macos (#4707) * Replace alias \Code with \TText (#4714) Fixed GH-2151 * Correct signing names and variables (#4713) * Add secrets to release workflow (#4719) * Add missing blosc2 info (#4717) * Fix error return types in H5Rdeprec.c (#4722) Copy-pasted code from elsewhere used FAIL instead of H5G_UNKNOWN and H5I_INVALID_HID. * Fix the release reference name (#4721) * Test creating unseekable file (#4720) * Cleanup up tests (#4724) * Add arch name to dmg file name (#4732) The binaries in snapshot dmg file do not work on x86_64. * Fix snapshot CI failure by adding arch name to dmg file (#4734) See also #4732. * Fix incorrect VOL vs. non-VOL calls partially (#4733) * Fix incorrect VOL vs. non-VOL calls H5Lget_info2() called H5I_object() instead of H5VL_vol_object() crashed user application. This is a wide-spread issue (GH-4730) but this PR only addresses GH-4705. * Remove an incorrect change * Fix segfault in ROS3 credential parsing (#4736) * Fix segfault in s3 credential parsing * Fix AWS cred parsing when >1 profile provided * Revert gh-pages action hash to fix daily build (#4735) * Revert gh-pages action hash to fix daily build See also #4734 * Revert gh-pages action hash to fix daily build * Eliminate another use of H5E_clear_stack() within the library (#4726) * Remove call to H5E_clear_stack() Also clean up a bunch of error macros and the return value from H5B_valid()
|
I just tried HDF5 v1.14.5 on Windows with Intel oneAPI 2024.2. It seems the error in #3663
etc. are still preset. Maybe these changes didn't make it into 1.14.5? |
This addresses the missing dependency for Windows of the shlwapi library, this will close #3663