Python 2.5 -> Python 2.6 [invalid]

kokosek 2010-11-12 11:42:14 UTC #1

Hi.
A user posted in task INTER an interesting comment.
Maybe he's right and Python 2.5 should be replaced by Python 2.6? Then, as he said, "the old 2.5- and 2.6-programs should run" (now only 2.5-programs works).

numerix 2010-11-17 17:01:01 UTC #2

Python 2.5 shouldn't be replaced by 2.6! It it correct that all programs coded for Python 2.5 will run with Python 2.6, but Python 2.6 is somewhat slower than Python 2.5 and thus you'll add more problems to the list of "TLE with Python"-problems. Many AC Python solutions would get TLE if rejudged. As Python 2.5 (especially in combination with psyco) is the fastest Python version ever, it should never be replaced!
Instead the second Python branch (actually with Python 3.1) should be updated to newer versions from time to time.

hendrik 2010-11-17 17:27:30 UTC #3

The evolution of Python especially in terms of performance isnt really good. This has been discussed earlier when 2.5 was replaced by several 2.6 versions. Whenever a new 2.6 version was installed it took more and more time to run. It became even worse with 3.X.
As it is all about time (except of course for the SHORTENING contest) I would also appreciate to not replace 2.5.

gorlum0_sof 2010-11-18 01:48:48 UTC #4

No replacement!
He can save bytes in the same way using python3. Time limit is forgiving enough here.
Wouldn't be hard to convert I guess.

As for evolution hope JITs would change things someday.)

numerix 2010-11-18 06:01:21 UTC #5

That's only true for some cases, mostly it isn't, if shortening is the target.

gorlum0_sof 2010-11-18 11:35:01 UTC #6

Hmm thought most of shortening comes from libs => 3 instead of 2.6 is ok.
But if not yeah. Anyway you know better.)

kokosek 2010-11-26 20:41:37 UTC #7

Thank You for presenting Your point of view!
Fortunately, I didn't manage to fool up. Phew!