Dare Obasanjo wrote:
> The whole "GETs must be side effect free" is a spec bug as far as I'm concerned. I can understand saying that they should be idempotent but besides opening the door to a bunch of metaphysical discussions aboutt "what exactly constitutes a side effect" I don't see what purpose that part of the spec serves.
Right. Basically every GET writes a logfile record as a minimum. I'm
more comfy with language along the lines of "GET does not incur new
obligations." Which is actually functionally necessary for caches and
proxies to work. -Tim
