Re: getmail failing for messages with invalid uncode

Geert Stappers <[email protected]> wrote:
> 
> another good reason to name "getmail6" without 'getmail'.

Indeed.  I'm hoping Roland does the right thing here.

> Who learnt from this thread that Python3 is NOT so good at unicode as
> advertised ...

Actually, Python 3 is a huge improvement.  Python 2 would implicitly and
silently convert between str and unicode (equivalent of bytes and str in
Python 3) in some cases, which would cause your code to work fine until you
happened to pass an encoded string containing characters with the high bit
set, which would then cause your application to explode.

Python 3 refuses to convert them implicitly.  You have to handle them
separately in your code, and explicitly encode and decode them when converting
between them (and specifying the correct encoding!).  It's much easier to make
sure your text/bytes ASCII/unicode handling are correct.

The problem is a lot of programmers - at least Western-world ones, and
particularly those working in English, where ASCII is a useful approximation
of "possible text" - don't really understand text, unicode, code points, and
encodings.  They tend to deal with it cargo-cult programming style, randomly
inserting `.encode(...)` and `.decode(...)` until they get something that
works.  It took me a few years to really "get" unicode handling.

Charles
-- 
-----------------------------------------------------------------------
Charles Cazabon
GPL'ed software available at:               http://pyropus.ca/software/
-----------------------------------------------------------------------