Kai Mühlbauer

@hmaarrfk with h5py 3.7.0 out our tracking test: https://github.com/h5netcdf/h5netcdf/blob/9c0269fa78286dd5a3547a040dd3b6c8902e9a68/h5netcdf/tests/test_h5netcdf.py#L1597-L1614 isn't failing any more. We would need to reevaluate the situation.

see https://github.com/h5netcdf/h5netcdf/pull/173

Looks like we can reinstate `track_feature=True` as default.

@shoyer Regarding reading file-like objects, is this still an issue? And second question, should we try to implement this in near future?

> And second question, should we try to implement this in near future? @shoyer I wasn't clear with my question. What I wanted to ask is, should we try to...

> I'll reopen this in case someone would like to give it a shot. Your message was first shown after I sent mine. I can give it a try, but...

This is like a phone call via a moon relay :grinning:

Thanks for raising this. I've thought about that recently, as this annoyingly has also other side-effects. If you take the following example: ```python f.create_variable("x", ("x",), data=np.ones((10)), ``` It creates the...

I've tried to untangle this in #101 but it was heavily involved in everything. So I took the detour via all the other PR's changing only small parts. The latest...

Let's assume this a bug, so we need to fix it.