I'm looking at the stale-* options again and wondering if we have any good
reason to port the update_headers option that is part of that feature? It
provides an HTTP violation.

I think the disk->memory promotion we already do for HITs deprecates this
as a performance boost.

Amos

Reply via email to