MERGE ping

[Aaron Bentley]

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

Robert Collins wrote:
> On Sat, 2006-03-25 at 11:34 -0500, Aaron Bentley wrote:
> 
> win32:
> 
> builtins.py - this is an improvement but not quite right yet - we should
> not need a write lock at all unless we are setting parent, and theres no
> point in setting parent if remotebranch == localbranch. +1 as is, would
> be nice to fit in with my patch [sent separately] to not take a write
> lock at all.

Okay.  Let's do it in stages.

> config.py - not sure why this is needed, but +1.

There are tests that assume unix line endings, and ConfigObj will use
native line endings unless a file object is supplied.  (Since then, a
new version of ConfigObj has been released that reportedly fixes this.)

> the transport.txt api tests - I wonder if we should have an osutils
> getcwd that is unicode and returns normalised ('\\' -> '/') paths
> always. +1 though, +2 if we do the osutils thing. 

We have one, but I had thought it would be bad to introduce more
internal functions code that's meant to be an example.  On reflection, I
think my changes are less readable than using the osutils function.

> the local.py change - perhaps changing the open clause to 'ab+' might
> help ? That might hint that reading and thus tell() having good data is
> needed.

I'll get back to you on this.

Aaron
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.1 (GNU/Linux)
Comment: Using GnuPG with Thunderbird - http://enigmail.mozdev.org

iD8DBQFEKBKs0F+nu1YWqI0RAjEsAJ0Y7h6G7YB/72ApO+AGnyDmPzlHXQCffEBZ
w+CLa3BEvBuTpnAk5D/BHiQ=
=oouV
-----END PGP SIGNATURE-----