Closed ghost closed 9 years ago
It's hard to sort out the relevant change among all the white space updates. Is just around line 1066 that has the relevant update?
I apologize for the "white noise" of changes. Can you configure your text editor to remove all the excess white space?
I put "Here!" to mark relevant code changes.
I just checked h5py's tests and fixed-length Unicode datasets are not supported. Vlen Unicode datasets are.
test script passes ok, went ahead and merged to master.
These changes fix this issue: HDFGroup/hpd-ws#26.
I traced the problem to assignment of a Unicode string object, e.g.
u"Hello world!"
to an already created dataset of the ASCII datatype (per its JSON information). When the json package converts JSON to Python dictionary all strings, its keys and string values, are Unicode string objects.The crux of the changes are:
I would keep this branch alive for testing when assigning str or unicode objects to UTF-8 string datasets.