Re: [Evolution] Filing system

[NotZed <notzed helixcode com>]

> "Bruce C. Dillahunty" wrote:
>
> > I agree with all of this... I have also been looking at things in this
> > light... one product that may pkay in this space (admittedly a commercial
> > product) is Oracle's iFS. Another is reiserfs... right now they are
> > concentrating on the journalling file system, but
> >
> > http://devlinux.com/projects/reiserfs/whitepaper.html
>
> Very interesting. Thanks for the link.
>
> BTW, it surprises me that none of the evolution developers have anything
> to say on the topic. Is this the wrong list? Should I be sending this to
> evolution-hackers?

Well, we've been a little busy, and not all of it seems particularly
relevant (talking about object filesystems for example), especially
since everyone will have their own ideas, some of which fall more
into the realm of research than feasible implementations.

Right now no virtual folder implementation even exists (i would personally
like to start looking at it fairly soon - mbox is sort of
getting usable), and i plan it to be built on searches inside
of existing mail stores.

So infact, the actual storage and search mechanisms used in a given
folder implementation will be transparent (and rather irrelevant)
to the virtual folder mechanism, so long as they all support the
same search semantics.  This issue covers most of the implementation
of the virtual folder.

What would probably be more constructive, is how do people think
the interface to such a system would work?  Sure, you have some
mechanism (again, the details are important, but irrelevant in
terms of the implementation) to define some search pattern to
apply to your folders, and that gives you back a result - what
happens if you delete a message (it deletes it from the source
folder?)  What happens if there are duplicate messages in different
folders returned by the same query?  What happens when you drag
a message from one vfolder into another?

 Michael