Currently, the Feats column is defined as – essentially – a dict of string → string. So far, this was quite OK, perhaps with the exception of Loanword, which is initially stored as a boolean, but serializes as "True" or "False" and is read back as a string.
However, @vikibrezinova is adding an Etymology feature, which is conceptually a list. It currently serializes as e.g. Etymology=['něm.', 'lat.', 'ř.'], which is ugly, and it deserializes as a single string.
There are two possible solutions:
- Force all
Feats values to be strings, therefore storing the etymology list as a single value. It would be simple, but working with the list would be cumbersome.
- Add some type information somewhere which would be referenced both when serializing and deserializing and which would ensure correct representation. Here, we probably want something like
Etymology=něm.,lat.,ř.
I vote for option No. 2.
Currently, the
Featscolumn is defined as – essentially – a dict of string → string. So far, this was quite OK, perhaps with the exception ofLoanword, which is initially stored as a boolean, but serializes as"True"or"False"and is read back as a string.However, @vikibrezinova is adding an
Etymologyfeature, which is conceptually a list. It currently serializes as e.g.Etymology=['něm.', 'lat.', 'ř.'], which is ugly, and it deserializes as a single string.There are two possible solutions:
Featsvalues to be strings, therefore storing the etymology list as a single value. It would be simple, but working with the list would be cumbersome.Etymology=něm.,lat.,ř.I vote for option No. 2.