[pycrypto] Once again: Python3 with PyCrypto

[Thorsten Behrens]

On 12/27/2010 8:37 AM, Dwayne C. Litzenberger wrote:
> Wow, it looks like you're really moving things along.

I am. Right now, the tests run and fail. Going through the first batch, 
it looks like it's all because of string/bytes things that need to be fixed.

I can use your input here, as I have two options with the way I resolve 
these issues:

Option 1:
Use helper functions instead of built-ins. Helper functions are used 
where binary data was intended. For example, ord(x) becomes bord(x), 
which returns ord(x) on 2.x and x on 3.x

Advantage: Explicit control. Where ord(x) on a text string was truly 
meant, it can still be called. Nothing unexpected will happen.
Disadvantage: Need to use the helper functions where before you could 
use a built-in

Option 2:
Assign ord = int if Python 3.x is detected.

Advantage: No code changes for ord, things just work.
Disadvantage: Potential side-effects that would be very hard to track 
down. Also, wouldn't be consistent with how we need to handle other 
things such as literals, where b('x') is the only option.


I am favoring Option 1, heavily. But before I change the code 
accordingly, I'd like to know you're good with the proposal.

Thorsten