Backup my script to generate result summaries

Perhaps it'll help someone else, some day.

documentation/build-results.sh

+#!/bin/sh
+#
+# Hacked-together script for building the results summaries on
+#  http://people.freedesktop.org/~nh/piglit/results/
+# The script is based on a database of results in the ${databasedir}.
+# The ${classesdir} is supposed to contain one file classes which
+# is simply a list of class names, and one file for each class which
+# contains names of results for that class, in chronological order.
+#
+# So for example, ${classesdir} might contain the files:
+#
+# ${classesdir}/classes:
+#   R300
+#   R500
+#
+# ${classesdir}/R300:
+#   R300ND-2008-06-12
+#   R300ND-2008-06-13-2
+#   R300ND-2008-06-30
+#   R300ND-2008-07-04
+#
+# And similarly, ${classesdir}/R500.
+# Corresponding results of test runs must exist in ${databasedir}
+
+
+### Configuration
+piglitdir=~/dev/xorg/piglit/repo
+resultsdir=../html/piglit/results
+classesdir=./classes
+databasedir=./database
+
+
+### Script starts here
+SUMMARY=${piglitdir}/piglit-summary-html.py
+
+mkdir -p ${resultsdir}
+classes=$(cat ${classesdir}/classes)
+
+echo "[" > tmpresults.all
+
+latest=
+for class in ${classes}; do
+	echo "Building report for class ${class}"
+	echo "['${databasedir}/$(tail -n 1 ${classesdir}/${class})'," >> tmpresults.all
+	echo "{'name': '${class}', 'href': '../${class}/index.html'}]," >> tmpresults.all
+
+	classlist=$(tail -n 4 ${classesdir}/${class} | sed s@^@${databasedir}/@)
+	${SUMMARY} -o ${resultsdir}/${class} ${classlist}
+done
+
+echo "Building report across classes"
+echo "]" >> tmpresults.all
+${SUMMARY} -o -l tmpresults.all ${resultsdir}/all
+rm tmpresults.all