[BUG] bzr 0.8 add on Unicode filename fails

[John Arbash Meinel]

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

John Whitley wrote:
>>> John A Meinel wrote:
>>>> I have a branch which has done a lot of the work, available here:
>>>> http://bzr.arbash-meinel.com/branches/bzr/encoding/
> 
> I'm still seeing this issue (reported as [1]) on Cygwin with bzr.dev
> revno 1816.  bzr add does work correctly on Mac OS X now (modulo the
> unicode normalization issues cited by John previously).  The exception
> trace is included below.
> 
> -- John
> 
> [1] https://launchpad.net/products/bzr/+bug/43689

This is a bug in cygwin python.
The problem is that cygwin python only supports ASCII as the locale.

So no matter what, extended characters won't work.

You probably would be able to use win32 python to handle those files.
But I don't think there is much that we can do for cygwin.

For example, this fails under cygwin, but succeeds under win32:

open(u'\xe5', 'wb').write('foo')

Now, the specific bug you are seeing might be fixable, but long-term
supporting unicode characters through cygwin doesn't work.

It also isn't limited to the python tools. If you have a unicode
filename, 'rm' doesn't work.
Which also means that if you run the test suite under win32 python, and
it fails to clean up, you can't just do rm -rf test000?.tmp. (I run
cygwin to get bash as a much better shell than cmd.exe, but I run win32
bzr, because it is much faster.)

You can open a bug if you want, but I think it is a cygwin issue that we
can't do much about.

John
=:->

-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.2.1 (Cygwin)
Comment: Using GnuPG with Mozilla - http://enigmail.mozdev.org

iD8DBQFEoXRjJdeBCYSNAAMRAkwrAJ4ySSrL7tL9+01LG5j0bYd83P2neQCdFoMn
JMvMvx6S3/y4jpTrIq4J06Q=
=sJeJ
-----END PGP SIGNATURE-----