I have spotted the biggest bottleneck in "bdiff.c". Actually it was
pretty easy to find after I recompiled the python interpreter and
mercurial for profiling.
In "bdiff.c" function "equatelines" allocates the minimum hash table
size, which can lead to tons of collisions. I introduced an
"overcommit" factor of 16, this is, I allocate 16 times more memory
than the minimum value. Overcommiting 128 times does not improve the
performance over the 16-times case.
#!/bin/sh
hg clone http://localhost:20059/ copy
echo $?
test -d copy || echo copy: No such file or directory
cat > dumb.py <<EOF
import BaseHTTPServer, SimpleHTTPServer, signal
def run(server_class=BaseHTTPServer.HTTPServer,
handler_class=SimpleHTTPServer.SimpleHTTPRequestHandler):
server_address = ('localhost', 20059)
httpd = server_class(server_address, handler_class)
httpd.serve_forever()
signal.signal(signal.SIGTERM, lambda x: sys.exit(0))
run()
EOF
python dumb.py 2>/dev/null &
echo $! >> $DAEMON_PIDS
# give the server some time to start running
sleep 1
http_proxy= hg clone http://localhost:20059/foo copy2 2>&1 | \
sed -e 's/404.*/404/' -e 's/Date:.*/Date:/'
echo $?
kill $!