|
Boost : |
From: Beman Dawes (bdawes_at_[hidden])
Date: 2003-06-12 17:27:10
At 01:31 PM 6/12/2003, Daryle Walker wrote:
>On Thursday, June 12, 2003, at 11:23 AM, Beman Dawes wrote:
>
>> An updated version of Boost.Filesystem has been added to CVS. The
>> primary change is adding "." as a directory-placeholder to the generic
>> path grammar.
>>
>> This solves the problem of distinguishing between an empty path and
>> one that acts as a placeholder.
>>
>> This change does change some existing behavior: path("foo/..") used to
>> get converted to an empty path. Now it gets converted to a path of
>> ".", and thus the change will break any existing code which depended
>> on the prior behavior.
>
>Is it possible to override this and use "." or ".." as regular object
>names?
No. Likewise there is no escape mechanism that allows you to include "/" in
a name.
There are a bunch of reasons - but particularly it would be creating names
that will just be rejected by many (or even most) modern operating systems.
What would be the point of that? It is the same as with requests for
allowing full URI syntax in paths; without any mechanism in the operational
functions allowing those paths, what would be the point?
--Beman
Boost list run by bdawes at acm.org, gregod at cs.rpi.edu, cpdaniel at pacbell.net, john at johnmaddock.co.uk