Skip to content

Conversation

@eschnett
Copy link
Contributor

This adds support for HDF5_jll version 2.0.0. There are a few minor incompatibilities that are corrected in this PR.

HDF5 2.0 adds support for complex numbers. This PR does not add such support. I assume adding such support is similar to adding support for Float16 (for which there is a separate PR).

One of the tests breaks; it seems something related to creating HDF5 files without tracking object creation times is not working.

@eschnett
Copy link
Contributor Author

Waiting for #1214 before continuing here.

@mkitti
Copy link
Member

mkitti commented Dec 10, 2025

Please let me know the status of this branch.

@eschnett
Copy link
Contributor Author

This branch is ready in principle. I do not know why the CI tests fail, I will investigate.

@mkitti
Copy link
Member

mkitti commented Dec 10, 2025

    [1] h5p_get_external(plist::HDF5.DatasetCreateProperties, idx::Int32, name_size::Int32, name::Vector{UInt8}, offset::Base.RefValue{Int32}, size::Base.RefValue{UInt64})

Looks like a another 32-bit to 64-bit issue

@eschnett
Copy link
Contributor Author

I don't understand the Julia 1.9 macOS error. One guess would be that there is no HDF5_jll artifact for this operating system and library. Maybe this uses libgfortran4, or maybe there is no respective MPI implementation available? These would all cause HDF5_jll to be quite empty, and the global constant there pointing to the dynamic HDF5 library would not exist.

If that's the case, I'm tempted to simply require Julia 1.10 as a remedy...

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

3 participants