[MERGE] Win32 fixes

[John Arbash Meinel]

Aaron Bentley wrote:
> John Arbash Meinel wrote:
>>> There are a *lot* more illegal characters than just '<>'. The list I
>>> always remember is '\/?<>":*|'
> 
> Sure.  My primary motivation was justifying skipping tests that used '<>'.
> 
>>> So we either need to treat ':' specially (as in make sure it is in the
>>> beginning, but doesn't exist in the rest of the path). Or just punt for now.
> 
> So would this be a good way of detecting a *legal* windows path?
> 
> re.match('([A-Z]:\\\\)?[^:<>*\"\'?]*$', 'C:\\ee')
> 
>>> But since baz-import uses 'Arch:foo' it would be a lot nicer for the
>>> user to get an IllegalPath exception, than some random OS error. 
> 
> Yes, it's pretty random, alright.  It looks like ENOENT!  Why oh why
> can't they give us a reasonable errno, so we don't have to predict
> what's valid?
> 
> Aaron

Actually, ' is a valid character on a windows path. And I'm sure we want
to support forward slashes instead of requiring back slashes. I thought
internally all paths were forward slashed anyway. And at the point we
are at, the path is an absolute path, so I would change it to:

_validWin32PathRE = re.compile(r'^[A-Z]:/[^:<>*"?\|\\]*$')
If we want to support either slash form, then it would be
_validWin32PathRE = re.compile(r'^[A-Z]:[/\\][^:<>*"?\|]*$')

John
=:->

-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 254 bytes
Desc: OpenPGP digital signature
Url : https://lists.ubuntu.com/archives/bazaar/attachments/20060421/1544406b/attachment.pgp