plug at plug slowness [was: Re: [plug] python head+wall issue]

[James Devenish]

Oh hi, I forgot about this thread...

In message <20031229135024.GA1626 at erdos.home>
on Mon, Dec 29, 2003 at 09:50:24PM +0800, Cameron Patrick wrote:
> A pretty graph can be found at: http://cp.yi.org/cameron/sparkdelay.png
[...]
> The interesting bit is that the delay looks like it increases and then
> drops back at the end of each month - perhaps it's archive-related?

I graphed my results, too, and the slowdown was definitely related to
Mailman, and surely to its archiving. Another odd feature is that the
speed was *improving* as time progressed toward mid-August. I notice
that our list activity has been fairly steady through the year, though
it seems to be greater in the last three months. I wonder if the list
server should be configured not to generate the archives for each
individual post but to generate the archive from a cronjob once every
30min? <http://www.guild.uwa.edu.au/~devenish/tmp/sparkdelay2.png>
Since I have been reading the list continually each day since
September, I think that accounts for why the outliers have had a
significant perceptual impact since that time.

In message <20031229153725.GA903 at erdos.home>
on Mon, Dec 29, 2003 at 11:37:25PM +0800, Cameron Patrick wrote:
> Okay, I'm curious: why is Octave good but gnuplot bad?
> Do you have a better alternative to suggest?

Maybe just a continuation of that list of stylistic preferences
you started:

you me
bash zsh
PC non-PC
rsync unison
gnuplot grace/xgobi