Re: [Gnu-arch-users] arch_inventory_traversal

[Tom Lord]

    > From: Florian Weimer <address@hidden>

    > Tom Lord wrote:

    > > There certainly are things that can happen that will change the
    > > inventory between calls to arch_inventory_traversal.   In other words,
    > > the _wrong_ change to make would be to modify arch_inventory_traversal
    > > to simply cache its results between calls.

    > If tla is hard-link safe, any change updates the mtime of a directory.
    > Therefore, it is sufficient to have a list of directory signatures to
    > check which directories have been updated and must be rescanned.

    > (You need a bit of fiddling to cope with the limited timestamp
    > resolution, but it's doable.)


Hey, neat.... that's quite true.

It's a little troubling to imagine a _clean_ implementation of that
vs. a quick one.....   

It occurs to me that a VU handler layer could be imposed that would
simply cause any non-hard-link-safe operation to fail, and
simultaneously optimize stat and readdir -- that would accomplish this
optimization with no changes to tla (other than installing the
handler).  (E.g., to prevent tla code from _ever_ opening a file for
write access other than with O_CREATE | O_EXCL).

The interaction with a hook to fire up an interactive log-file editor
is a little scary to contemplate.....


-t