Bug with imported Cmake target 1.14.0

[larsoner]

Solution to issue cannot be found in the documentation.

• [X] I checked the documentation.

Issue

The error in the automated OpenMEEG rebuild PR https://github.com/conda-forge/openmeeg-feedstock/pull/42 seems to indicate a problem with HDF5 feedstock:

https://dev.azure.com/conda-forge/feedstock-builds/_build/results?buildId=654485&view=logs&j=d0d954b5-f111-5dc4-4d76-03b6c9d0cf7e&t=841356e0-85bb-57d8-dbbc-852e683d1642&l=271

Running CMAKE
-- The C compiler identification is GNU 11.3.0
...
-- Found HDF5: $PREFIX/lib/libhdf5.so;$PREFIX/lib/libcrypto.so;$PREFIX/lib/libcurl.so;$BUILD_PREFIX/x86_64-conda-linux-gnu/sysroot/usr/lib/librt.so;$BUILD_PREFIX/x86_64-conda-linux-gnu/sysroot/usr/lib/libpthread.so;$PREFIX/lib/libsz.so;$PREFIX/lib/libz.so;$BUILD_PREFIX/x86_64-conda-linux-gnu/sysroot/usr/lib/libdl.so;$BUILD_PREFIX/x86_64-conda-linux-gnu/sysroot/usr/lib/libm.so (found version "1.14.0")  
...
-- Configuring done
CMake Error in OpenMEEGMaths/CMakeLists.txt:
  Imported target "HDF5::HDF5" includes non-existent path

    "/home/conda/feedstock_root/build_artifacts/hdf5_1675740981212/work/src/H5FDsubfiling"

  in its INTERFACE_INCLUDE_DIRECTORIES.  Possible reasons include:

  * The path was deleted, renamed, or moved to another location.

  * An install or uninstall procedure did not complete successfully.

  * The installation package was faulty and references files it does not
  provide.

Note that my manual migration PR to 1.12.2 went smoothly and we haven't changed OpenMEEG infra/code at all, so it seems like a new problem with 1.14.0.

Installed packages

Copied from https://dev.azure.com/conda-forge/feedstock-builds/_build/results?buildId=654485&view=logs&j=d0d954b5-f111-5dc4-4d76-03b6c9d0cf7e&t=841356e0-85bb-57d8-dbbc-852e683d1642&l=271

    _libgcc_mutex:            0.1-conda_forge         conda-forge
    _openmp_mutex:            4.5-2_gnu               conda-forge
    binutils_impl_linux-64:   2.39-he00db2b_1         conda-forge
    binutils_linux-64:        2.39-h5fc0e48_11        conda-forge
    bzip2:                    1.0.8-h7f98852_4        conda-forge
    c-ares:                   1.18.1-h7f98852_0       conda-forge
    ca-certificates:          2022.12.7-ha878542_0    conda-forge
    cmake:                    3.25.2-h077f3f9_0       conda-forge
    expat:                    2.5.0-h27087fc_0        conda-forge
    gcc_impl_linux-64:        11.3.0-hab1b70f_19      conda-forge
    gcc_linux-64:             11.3.0-he6f903b_11      conda-forge
    gfortran_impl_linux-64:   11.3.0-he34c6f7_19      conda-forge
    gfortran_linux-64:        11.3.0-h3c55166_11      conda-forge
    gxx_impl_linux-64:        11.3.0-hab1b70f_19      conda-forge
    gxx_linux-64:             11.3.0-hc203a17_11      conda-forge
    kernel-headers_linux-64:  2.6.32-he073ed8_15      conda-forge
    keyutils:                 1.6.1-h166bdaf_0        conda-forge
    krb5:                     1.20.1-h81ceb04_0       conda-forge
    ld_impl_linux-64:         2.39-hcc3a1bd_1         conda-forge
    libcurl:                  7.87.0-hdc1c0ab_0       conda-forge
    libedit:                  3.1.20191231-he28a2e2_2 conda-forge
    libev:                    4.33-h516909a_1         conda-forge
    libgcc-devel_linux-64:    11.3.0-h210ce93_19      conda-forge

### Environment info

```shell
Standard conda-forge build, see https://dev.azure.com/conda-forge/feedstock-builds/_build/results?buildId=654485&view=logs&j=d0d954b5-f111-5dc4-4d76-03b6c9d0cf7e&t=841356e0-85bb-57d8-dbbc-852e683d1642&l=271

[hmaarrfk]

Thanks. I would like to say that this is an upstream bug with malformed cmake file.

[hmaarrfk]

will need more investigation.

[ndevenish]

We've just hit this on the autoupdate PR on our feedstock https://github.com/conda-forge/dxtbx-feedstock/pull/31

[hmaarrfk]

I think homebrew hit the same bug https://github.com/HDFGroup/hdf5/issues/2422

[larsoner]

Awesome, thanks for investigating @hmaarrfk !

[hmaarrfk]

i'm somewhat tempted to pause the HDF5 migration. Though I'm not sure i fully want to stop it. I feel like without "trailblazing" nobody will ever test the new software, so we will always "delay" these tests.

I am building both versions of HDF5 for opencv at the moment thought.

[larsoner]

It would be nice if all or most conda packages that depend on the hdf5 version at least build though. I'm not sure how many use the cmake interfaces -- if 99% work without needing cmake that's one thing, if only 20% do that's another. So maybe 1.14.0 already works for most packages and the others can wait for 1.14.1? 🤷

I'm also not sure how much extra complexity and solve time having only some packages migrate would create, which might be a concern.

[hmaarrfk]

i think it is going "better than other HDF5 migrations". Lets see how fare we go.

[MicK7]

I see the same cmake issue in https://github.com/conda-forge/cgns-feedstock/pull/27

Thanks @hmaarrfk for investigating

[MichaelSchoenfelder-SiV]

This appears to be what is causing the error in my environment. Note the extra include path; that's what is causing the error since it doesn't exist.

h5c++ -showconfig | grep AM_CPPFLAGS
                    AM_CPPFLAGS:  -I<my conda env path>/include -I/home/conda/feedstock_root/build_artifacts/hdf5_1675775987664/work/src/H5FDsubfiling

[hmaarrfk]

https://github.com/HDFGroup/hdf5/issues/2422 was recently closed. If somebody wants to look into backporting the fix that may be appropriate while a new release is cut.

[minrk]

Fixed in #202