hdf5

hdf5 copied to clipboard

---

Various fixes from VTK's bundled HDF5 that we've applied that seems relevant for upstream.

I can split this into multiple PRs if that is preferred as well.

The proposed changes need justification/discussion

Please see the commit messages. Are there any that are deficient?

How about breaking the isnan change into a separate PR and also the H5S changes into another PR?

How about breaking the isnan change into a separate PR and also the H5S changes into another PR?

I agree. Smaller, more focused PRs are easier to review and merge downstream to other maintenance branches.

See #1708 and #1709 for those two split out.

@mathstuf Apologies, I made a bunch of changes recently to cleanup some of the remaining warnings in the develop branch and it seems it caused conflicts among your PRs; it might be worth rebasing them on the current develop. I fixed the H5S warnings, but in a different way. For dt_arith, I just replaced the direct self comparison with a comparison using an epsilon, but using isnan probably makes more sense.

Thanks; I'll get around to it on my next HDF5 update cycle.

Oh, there's stuff to perserve here.

Yeah, I've been meaning to go over this again to see what's up with all the changes.

I think we've pulled everything we can out of this. The one thing that is left is the cross-compiling bits, but I'm going to leave that to a future discussion.