Re: [dev] Recent vain attempts at suckless Web applications

From: Kurt H Maier <>
Date: Sat, 22 Jan 2011 17:58:28 -0500

On Sat, Jan 22, 2011 at 2:41 PM, Joseph Xu <> wrote:
> When we're talking about a small script with trivial running time
> (which is not necessarily the case here), the 10x slower startup time
> for python does make a difference when you're programmatically
> executing the script several thousand times.

...and this is precisely the garden path that people followed to
create modern massive web frameworks. "We'll just have an always-on
python program that directly serves http! Wait, then we can't use our
existing web daemon's caching, authentication, or load-balancing. No
problem! We'll just have multiple python httpds, sitting behind a
caching mechanism, and we'll proxy to that from a web server."

It's possible to get very good performance with such a stack ... but
it's definitely not in line with the methodology.

# Kurt H Maier
Received on Sat Jan 22 2011 - 23:58:28 CET

This archive was generated by hypermail 2.2.0 : Sun Jan 23 2011 - 00:00:05 CET