The pandas writers / reader is completely ignoring the string/symbol types.
-
When writing : both string and numpy.string are going to be passed as symbol when writing. But according to documentation only numpy.string should be converted to symbol
-
Similarly when reading: symbol/string are going to be received as bytes instead of numpy.bytes for symbol.
pandas_parser
possible cause - _pandas.py/line 165 :
if qtype == QSTRING: # assume we have a generic list of strings -> force representation as symbol list qtype = QSYMBOL