[Chicago] Kickstarter Fund to get rid of the GIL

Tal Liron tal.liron at threecrickets.com
Mon Jul 25 22:37:23 CEST 2011


One more link of interest by Armin Rigo, re: discussing GIL in various 
Python runtimes (yes, PyPy still has a GIL and will have one for a very 
long time):


http://morepypy.blogspot.com/2011/06/global-interpreter-lock-or-how-to-kill.html


If Beazley won't do it, you can pay PyPy to. :)


My opinion is that STM may not be the best fit for Python. As Rigo 
writes, everything in Python would have to be rewritten to implement 
STM, and "everything" really means everything.


STM is used to tremendous effect in Clojure, but it's an explicit 
feature of the language that you turn on when transactions make sense. 
Since Python has no such linguistic concept (nor the persistent data 
structures the make it so natural, it would mean transactions 
everywhere, all the time, behind the scenes. I agree with Rigo that this 
might end up being worse than the GIL.


-Tal



More information about the Chicago mailing list