Re: [Gnu-arch-users] Re: arch with 'special files'

[tomas]

On Sat, Apr 02, 2005 at 01:17:46PM +0200, Ulf Ochsenfahrt wrote:
> On Fri, 2005-04-01 at 08:07 +0200, Robert Widhopf-Fenk wrote:
> > On Wednesday, March 30, 2005 at 12:19:39, address@hidden wrote:
> > > On Tue, Mar 29, 2005 at 11:39:04AM -0600, John Meinel wrote:
> > > > 
> > > > >On Tue, 2005-03-29 at 11:01 +0200, Matthieu Moy wrote:
> > > 
[...]
> > > > >I'm pretty sure all the changes I'd like can be handled with more
> > > > >(optional) metadata.  I'm not against some scripting glue, but to
> > > > >do this I still need to be able to store/retrieve some metadata
> > > > >in the archive.
> > > 
> > > Heh. The metadata discussion again :-)
> > > 
> > > > If you are asking for user-defined meta-data, how is this
> > > > different from creating a user-defined text file listing the
> > > > metadata that you are keeping track of [...]
> > > 
> > > Well, it ain't different -- and it is. If Arch provides a
> > > standardized repository for (generic) file metadata, it's gently
> > > forcing applications to agree on one mechanism. 
> > 
> > And there would be no need to externally care for move and
> > remove of the metatdata along with a file [...]

[...]

> When talking about meta-data, you don't by any chance mean the file
> attributes that are already available for ext2, ext3, reiserfs and other
> filesystems.

Well... yes and no. Tla (any system archiving --hm... let's call it files--
tries to be file-system agnostic. Then one's `native´ metadata becomes
other's `alien´ metadata. See for example Unix permissions on DOS. Or
HTs EAs on ext2. Or the file attributes you mentioned on VFAT. Or one
of Macs funny forks. And it goes on. Why shouldn't one of Archs applications
define its own thingie it wants attached to some files (like, as has already
come up here: ``this file is an MSDOS text file and I want it be CRLF
whenever checked out on MSDOS and just LF when checked out on UNIX´´.

Or whatever.

The question being thrown back and forth here is: should Arch support this?
Does it make sense? If yes: at what level? Should it special-case some
``important´´ instances, like permissions (and maybe ownerships) and leave
the rest to users?

That's at least how I see this discussion (from my very limited POV).

Regards
-- tomás

pgpDLYafK9wjw.pgp
Description: PGP signature