visualize.parameters produces error when working with hdf5-stored result.problem.x_scales

[C-Peiter]

Seems like, when I load a result using result=pypesto.store.read_result("path/to/result.hdf5") the result.problem.x_scales attribute gets loaded as an array of byte-strings. This produces a type error at visualize/parameters.py:128 when all scales are the same. The error is TypeError: can only concatenate str (not "bytes") to str and otherwise prints the scale as, for example, b'log10'.

The result file was saved with

pypesto.store.write_result(
    result=result,
    filename="path/to/result.hdf5",
    problem=True,
    optimize=True,
    profile=True,
    sample=True,
)

OS: Ubuntu 22.04
HDF5 lib on ubuntu: libhdf5-103-1 1.10.7+repack-4ubuntu2
Python version: 3.10.12
pyPESTO version: 0.5.4
h5py version: 3.12.1

[stephanmg]

@C-Peiter Could you provide the snippet which produces the error with the .hdf5 file? Then I'll try to reproduce this on my Ubuntu and have a look. (And possibly the error output)

[C-Peiter]

I attached a result stored in hdf5 and the code to load the result and plot the parameters. I also have a whole PEtab problem for that, if you want.
mwe.zip

[stephanmg]

I attached a result stored in hdf5 and the code to load the result and plot the parameters. I also have a whole PEtab problem for that, if you want. mwe.zip

Thanks, that should suffice.

[stephanmg]

Think a simple decode('utf-8') should be enough, realized in PR #1542 --- @C-Peiter I did not see your suggestions ;) Sorry.

The issue is in pypesto.store.read_from_hdf5 where x_scales is not decoded from the byte str to str after reading the .hdf5 file, fixed in PR #1542.