From: Nobuyoshi Nakada <nobu@...>
Date: 2008-04-16T12:00:18+09:00
Subject: Re: cross-platform1: st1.dev == st2.dev and st1.ino == st2.ino considered harmful

Hi,

At Wed, 16 Apr 2008 00:21:05 +0900,
Thomas Enebo wrote in [ruby-core:16393]:
> >>   It does not get rid of the second use in file_utils which is based 
> >> on an lstat.   So it may still be a good idea to perhaps have 
> >> FileStat.identical? so results of lstat may still be tested.  This 
> >> could get rid of fu_stat_identical_entry?.
> >
> > Sounds nice to me, though I'm not sure if it is technically possible 
> > for a FileStat to detect its identity on Windows.

I think it is reasonable to define FileStat.identical? only on
the platform providing ino, so that we can check with
"defined?" or "respond_to?".

> I am not a windows person so I cannot say, but I would think there is 
> some internal windows API that could perform an identity check.  
> Daniel?  Some windows user can perhaps shed more light on this?

No persistent identification is available on Windows.
File.identical? looks BY_HANDLE_FILE_INFORMATION, but they are
vaild only during the file handle is opened.

-- 
Nobu Nakada