Re: [Summary] Meta-data/filesystem-encapsulation

[Miles Egan <cullen best com>]

On Sat, 15 Aug 1998, Kevin Littlejohn wrote:

> I think it comes down to there is no way to keep metadata and not loose it
> if you attack the filesystem with non-GUI tools (eg. mv/cp) (and I'm assuming
> you _don't_ have the ability to LD_PRELOAD all the time here), so whatever
> approach is used has to recognise that and allow for resynchronisation and
> graceful error handling.  Probably also, it needs to be reasonably obvious
> to the user _how_ it breaks - nothing worse than doing something you think
> should work, only to find it doesn't.

This sounds like a real mess to me and I think we should think very
carefully before we try and implement a scheme like this.  It seems to me
that you'll have to change the filesystem to make something like this
really robust enough to be a benefit to the user.  Changing the filesystem
just isn't an option for Linux.

Miles