Sequences implementation (was Re: [Nmh-workers] A script for threading)

[Joel Reicher]

> On Tue, Feb 07, 2006 at 07:57:16AM +1100, Joel Reicher wrote:
> > [ .. ]
> > potentially many records (sequences) to update. Without getting too
> > whacky, about the only fast way I can think to do sequences is by having
> > them as a list of inodes.
> 
> Everytime I've seen inodes exposed at the user level,
> there are tears eventually.  Some filesystems inodes
> aren't stable, sometimes people move (copy+delete) the
> mail etc, etc.
> 
> So please not!

My instinctive hesitation is more than happy to defer to your experience. :)

By the way, if anyone happens to think that the 26 sequence limit should
be lifted, speak up. Perhaps the performance hit wouldn't be so bad.

Cheers,

        - Joel