Latest ASE giving problems

[danieleongari]

With the latest version of ASE (3.21.1) I get the following error when loading a standard CIF:

  File "/Users/danieleongari/opt/anaconda3/lib/python3.8/site-packages/ase/io/cif.py", line 499, in parse_block
    assert line.lower().startswith('data_')
AssertionError

the problem does not occur with my previous version of ASE (3.19.1), so maybe let's give a look or constrain the version.

[ltalirz]

we've constrained to ase<3.20 in aiida-lsmo already

at some point we'll need to figure out "who's right" (or whether there is a way to make the new ase more lenient)

[ltalirz]

from the release notes:

3.20.2:

• CIF writer now uses up-to-date variable definitions from the CIF standard instead of deprecated ones. Also, it no longer writes columns of dummy data that doesn’t depend on the atoms.

Perhaps the reason is that we are using variable definitions that are deprecated in the CIF standard.

[ltalirz]

Change in the source code: https://gitlab.com/ase/ase/-/merge_requests/2253/diffs
particularly affects space group symmetry symbols