On 08/03/2016 01:40, Chris Angelico wrote:
On Tue, Mar 8, 2016 at 12:33 PM, BartC <b...@freeuk.com> wrote:
Let me ask you a follow-on question first: how slow does a new Python
version have to be before even you would take notice?
Compared with 2.7, 3.4 above is spending nearly an extra ten seconds doing
.... what? I can't understand why someone just wouldn't care.
Performance matters, when you actually have something useful to
measure. Startup performance matters enormously if interpreter startup
is what you're doing a lot of (for example, the "feel" of Mercurial
depends heavily on Python startup performance). It matters not a whit
if your process keeps running for a long time, and handles many
requests (for example, a web server).
You need to be VERY clear about exactly what you're measuring. Are you
using the 'timeit' module to measure execution of one line of code?
Are you putting your code into a file and running that with
/usr/bin/time? Are you putting the code into Idle and running it in a
loop with 'exec' and using time.time() around the outside? Your
numbers do not concern me *because they mean nothing*.
So they're just random numbers?
I write interpreters. If I suddenly saw a 115% increase in runtime from
one version to another, /even on a trivial 2-line loop/ (or even on
/any/ program doing the same task), then I'd want to know why! Because
experience tells me that something is wrong.
Of course I don't know here what they've been doing to Python. Maybe
they know exactly why it's slower. But then they should explain why it
is, otherwise it could be a bug. Maybe someone left a debug flag turned
on or something. Or there's some other technical reason, but it would be
nice to know what it is.
--
Bartc
--
https://mail.python.org/mailman/listinfo/python-list
