Comments on: Making re-creatable random data files really fast in python.

By: jnoller

jnoller — Sat, 07 Jun 2008 16:52:38 +0000

For some reason (which I never got a chance to look into) calls into random are expensive/slow, which is why I did it this way.

By: Eric Brunson

Eric Brunson — Sat, 07 Jun 2008 16:32:00 +0000

Why not use a pseudo-random number generator and start with a known seed? Seeding Python's built in PRNG will result in the same sequence and is fast. Am I missing something?

By: jnoller

jnoller — Mon, 02 Jun 2008 17:21:53 +0000

Thanks, I'll give that a shot too

By: Anonymous

Anonymous — Sat, 31 May 2008 22:58:26 +0000

Hmm. Let's see if this works better:

for chunk in fdata():
fh.write(chunk)
size -= len(chunk)
if size < 0:
break

By: Sergey

Sergey — Sat, 31 May 2008 13:30:05 +0000

Nice! Unless I'm missing something, .replace(”n”, '').split() could be just .split(). Newline is one of the default separators: http://www.python.org/doc/current/lib/node42.ht....

By: Anonymous

Anonymous — Sat, 31 May 2008 06:25:37 +0000

The easiest way to eliminate the getsize call would be something like:

for chunk in fdata:
fh.write(chunk)
size -= len(chunk)
if size < 0:
break

By: Christian Wyglendowski

Christian Wyglendowski — Fri, 30 May 2008 22:35:53 +0000

Hey, that is cool. Can't think of where I'd need this right away, but still, very cool.