Hi Andréas,
The thing with python3 is it dropped implicit conversion between binary and "textual" strings. That means that:
- you now need to explicitly encode or decode if you need to switch the type of string you're using
- any conversion issue will be risen even if you're just using plain ascii charset
With python 2, encoding / decoding error where more random because the API were not always clear whether they need a string or a unicode and it depended on the string value itself.
In python3 this kind of error will happen right from the start.
While I've been porting things to python3, I've seen a couple of places where there was latent encoding/decoding errors that weren't identified before.
Also I love Python3 for being more explicit than Python 2.
Another example: import path now only use absolute path. If you want to use a relative import, you'll need to be explicit about it (say from . import models).
Best regards,
Xavier,
Linovia.