Comment 8 for bug 662523

it's not particularly small anymore.

see lp:~fta/+junk/ppa-dashboard.trunk

here is how i run it (from within the branch) to create what's visible there: http://people.ubuntu.com/~fta/ppa-dashboard/

=====================================
OUTDIR=$(mktemp -d --tmpdir=.)
BIN=./dashboard.py

OPTS="--dir $OUTDIR --cachedir cache"

D1=$(date +%s)
$BIN $OPTS -r --grouped --output chromium-daily.html chromium-daily/stable,beta chromium-daily/ppa,dev
$BIN $OPTS -r --grouped --output mozillateam-stable.html mozillateam/firefox-stable mozillateam/thunderbird-stable
$BIN $OPTS mozillateam/ffox35
$BIN $OPTS -r ubuntu-mozilla-daily/ppa
$BIN $OPTS -r ubuntu-mozilla-security/ppa
D2=$(date +%s)

DELTA=$(expr $D2 - $D1)
echo $(date +%Y/%m/%d-%H:%M:%S)" $DELTA sec" >> runtime.log
========================================

it should not output anything in the shell. Everything is in $OUTDIR.
the 1st run is slower (as it starts with an empty cache), next runs should be way faster.

Add -v tp $OPTS to see the perfs of individual LP calls.

Compared to the branch, i'm currently running with smaller chunks because of this bug:

=== modified file 'dashboard.py'
--- dashboard.py 2010-10-24 18:35:20 +0000
+++ dashboard.py 2010-10-26 10:06:27 +0000
@@ -348,7 +348,7 @@

     # Split the list of remaining source-ids into chunks of reasonable size to avoid timeouts.
     # ~25 per request should be okay
- sz = 25
+ sz = 10
     chunks = map(lambda i: new_sids[i:i+sz], xrange(0, len(new_sids), sz))
     for chunk in chunks:
       t = self.perf.start("getBuildSummariesForSourceIds(%d sids)" % len(chunk))