Need advices regarding the strings (str, unicode, coding) used asinterface for an external library.

[jmfauth]

I'm planning to build an external lib. This lib will exchange
a lot of strings between the lib and the "core Python code"
of applications.

I wish this lib to be modern, 100% unicode compliant. It will
be developped for Python 2.7 and for Python 3. In an early
phase, technically, it will be developed on Python 2.7 before
Python 3, probably 3.2.

Two options for the strings interface.

a) Pure unicode, that means only type 'unicode' in Python 2.7
and only type 'str' in Python 3.

Similar to the Python io module.

b) Like a) plus ascii and utf-8 encoded type 'str' to keep
some kind of retro compatibility. This lib will anyway
work in a "unicode mode", so the ascii and the encoded
utf-8 str's have to be converted into "unicode".

I'm very comfortable with all this coding stuff
and aware of the pros and cons of each solutions.

My favourite solution is clearly on the a) side.

Advices and comments are welcome. Thanks in advance.

 

[Terry Reedy]

I'm planning to build an external lib. This lib will exchange
a lot of strings between the lib and the "core Python code"
of applications.

Are you planning to exchange indirectly via disk files or directly via
memory buffers?

This pretty much amounts to whether the library will interface with
Python-only or with anything.

Also, what OSes? If 'all', you need to be able to work with both 2 and 4
byte unicodes.

 

[jmfauth]

Thanks for the reply. Subject closed.