A critic of Guido's blog on Python's lambda

Alex Martelli aleax at mac.com
Tue May 9 17:14:03 CEST 2006

Rob Warnock <rpw3 at rpw3.org> wrote:
> If "only" being useful is enough, 100 cycles is enough for a DNS server,
> or an NTP server, or even a stub HTTP server that delivers some small
> piece of real-time data, like a few realtime environmental sensors
> [temperature, voltages, etc.].

Reminds me of Stuart Cheshire's description of how they managed to
shoehorn zeroconf (aka bonjour, the artist formerly known as rendezvous)
into a risible amount of ROM (less than 1K byte, if I recall correctly)
left in an embedded microcontroller (for a video camera, I think).
Zeroconf is at heart a few clever tricks on top of DNS (plus 169.254.*
IPs), and in the end they managed by one more clever trick (the thingy
ignores WHAT the request is for, and just spits out the same response
each and every time -- pushing the boundaries of DNS but, it seems,
still technically staying within those boundaries;-).

Not directly relevant to the scaling debate (the camera's expected to be
on a LAN, serving a few requests per second at most), but the limit
being on bits rather than cycles somehow "tastes" similar to me.


More information about the Python-list mailing list