[sugar] Tagged Journal Proposal

[C. Scott Ananian]

On Tue, Sep 23, 2008 at 6:05 PM,  <pgf at laptop.org> wrote:
> c. scott ananian wrote:
>  > On Tue, Sep 23, 2008 at 5:57 PM, Eduardo H. Silva <hoboprimate at gmail.com> wrote:
>  > > Ah, so that's why you separate these legacy-hierarchical files with a
>  > > light grey slash (/) . So that a kid who only knows the Journal
>  > > tagging world can ignore it, and users who have know the hierarchical
>  > > world can understand it and make advance usage of that knowledge when
>  > > transfering from or browsing hierarchical filesystems.
>  >
>  > Exactly. =)
>
> seems like acknowledging the "path form" of these
> directory-derived tags might also make working with devices for
> which no tag list has been, or can be, created.  i.e., when you
> first install a large new USB stick, there will certainly be a
> delay before a tag index can or will be built.  the grey slashes
> might be black during that time.

Hah, you're getting into implementation details now.

I believe that anyone creating indexes for or on removable devices is
living in a state of sin.

The searches should still work immediately, with no indexes: they
might just take a while.   This ends up being a recursive directory
traversal, but it's not Death. That's fine, we can show the immediate
results immediately, and the rest just take a while.

We can probably write some hints for use during the current session,
esp for large devices (attached USB cards, persistent SD) but we can't
assume that these hints are persistent across mounts, or aren't
corrupted in the process of removing the device.  (As a special case,
for devices formatted ext2/3 we can probably play some tricks with the
mount count and dirty flag to let us know when it's safe to use old
hints.)
 --scott

-- 
 ( http://cscott.net/ )