Re: No os.copy or lfs.copy?

[David Dunham <david@...>]

On 10 Apr 2008, at 04:05, gary ng wrote:

> > This is pretty off-topic, but resource forks are not
> > metadata.
>
> off-topic or not, I am curious to know why resource
> forks are not metadata.

Because they are the actual data. Metadata is typically optional.

A Mac file can have BOTH data and resource forks, and BOTH can have  
data in them.

For example, the Mac Classic outliner Acta saved text in the data fork  
and sound annotations in the resource fork.

A blind copy would copy just the data fork, and thus strip out the  
recorded sound.

I know the Mac OS X "cp" command now deals with this, but I think this  
behavior was added only in 10.3 or so.

> Say if I have a 'file'(in modern day sense they are
> just a byte stream), anything that is not in this
> stream(which are not handled by read then write by the
> file module) are metadata, no ?


You're still making the assumption that a file is a byte stream. It is  
not. It's a collection of forks (the resource fork and the data fork  
can each be considered byte streams).

I'm not as familiar with the recent Windows file systems which have  
multiple forks, but I think they'd be a problem as well.

David Dunham       Development Manager
+1 206 926 5722      GameHouse Studios