Thanks, I read it. But I didn't see any conclusion about letting a version of python (2.5 or wherever it is) that has psyco support.
I did a small experiment with my python solutions and I got one worrying result. I have [color=#FF0000]48 AC problems[/color] in http://br.spoj.pl/status/diegocaminha/signedlist/ using python 2.5 + psyco. After the system update, I resubmitted my solutions just commenting 2 lines "#import psyco; #psyco.full()" and got to this:
[color=#FF0000]19 AC
29 TLE[/color]
That is only 39 % of my solutions are "fast enough" for python 2.6. And I know that most of this is just because of the input reading.
I'm not a python expert. I just use it here for fun and to practice a little bit, but I wonder if such update is worth when you limit so much the number of problems "solvable" 
