Interesting Read: Tear Down that GIL!

by jesse in ,


This is a post that popped up this AM for me - "An open letter to Guido van Rossum: Mr Rossum, tear down that GIL!". It's an interesting read if nothing for the fact that the GIL is the biggest piece of cane with which people beat python with (I don't count significant whitespace whining). Take a peek at that, and then read the reddit discussion. I've been writing a piece on the GIL off and on for a few weeks now - collecting discussions and information and trying to piece together the most cognizant series of thoughts I could muster.

The interesting thing is that the GIL was also one of the point Bruce Eckel touched on in his recent "Python 3K or Python 2.9?" post on the 8th - be sure to read the comments there too.

I've collected about 30 or so GIL/Concurrency related tidbits I'll put together in a piece, but I am interested in hearing other opinions about the GIL before I reveal my particular horse in this race.

Call me wierd - but threading itself is a bit of a beast in my eyes: it could be because I've never built a giant UI application, or I am used to talking amongst machines rather than cores on a single machine. It seems(?) better to plan to scale across nodes in a cluster rather than processors on the machine. Coding to the former begets the latter, no? Note that I do use threading - quite freqently I might add.

Update: Michael Tsai weighs in on the "open letter".

Update Part 2: Guido's posted a response to the open letter. I think this clarifies the BDFL's position as one of "give me something and we'll see". ( And the reddit discussion now )