From: Beman Dawes (bdawes_at_[hidden])
Date: 2003-08-14 19:49:06
At 03:28 PM 8/14/2003, David Abrahams wrote:
>"Peter Dimov" <pdimov_at_[hidden]> writes:
>> I am not sure that it should be the responsibility of the path class to
>> enforce some notion of portability. Wouldn't it be more appropriate to
>> defer the portability check, if any, to the point where the path is
>> actually used in a filesystem operation?
>I agree. Having portability enforcement can be useful, but AFAICT the
>way it's currently supplied is biased for the 1% case. There's an
>enormous class of applications which gets paths from the user or one
>part of the platform implementation (e.g. a file selection dialog),
>possibly does some manipulation on the path, and uses it to open some
>local files on the platform.
Yes, of course. There is no check performed in those cases. The only names
checked are on those portions of a path created via the generic grammar
constructors. The check does not apply to the native constructors.
> The only category of application which
>needs enforced path portability, AFAICT, is the one which stores paths
>in some file meant to be used on other platforms, or arguably those
>which encode paths directly in portable code. These applications are
>comparitively rare IME.
Only those cases are checked, so if they are rare in your code, then you
won't experience a problem. Let me know if you run into any cases where
name checking gets applied to a native path; that shouldn't be happening.
Boost list run by bdawes at acm.org, gregod at cs.rpi.edu, cpdaniel at pacbell.net, john at johnmaddock.co.uk