Fwd: Lossless bulletproof conversion to unicode (backslashing)

[anatoly techtonik]

On Wed, May 27, 2015 at 2:35 PM, Laura Creighton <lac at openend.se> wrote:
> I am missing something.  Why do you need unicode at all?  Why can you
> not just keep your binary data as binary data?

Good question. From the SCons code I see that we need unicode, because
we switched to io.StringIO which is advertised as the future (and Python 3
way of doing things, because Python 3 doesn't have non-unicode StringIO).

A really deep and exhaustive answer.
advertisement (first link on StringIO vs io.StringIO):
http://stackoverflow.com/questions/3410309/what-is-the-difference-between-stringio-and-io-stringio-in-python2-7
peaceful details
https://bitbucket.org/scons/scons/commits/05d5af305a5d
gory consequences
https://bitbucket.org/scons/scons/pull-request/235/fix-tree-all-print-when-build-tree

> I feel like I must be missing something obvious here ...

Not that obvious as it appears.
-- 
anatoly t.