Le Tue, 09 Mar 2010 15:56:47 -0800, Metalone a écrit : > for i in xrange(1000000): > f1.seek(0)
This is quite a stupid benchmark to write, since repeatedly seeking to 0 is a no-op. I haven't re-read the file object code recently, but chances are that the Python file object has its own bookkeeping which adds a bit of execution time. But I would suggest measuring the performance of *actual* seeks to different file offsets, before handwaving about the supposed "slowness" of file seeks in Python. Regards Antoine. -- http://mail.python.org/mailman/listinfo/python-list