Interesting Read: Tear Down that GIL!

September 10th, 2007 § 0 comments

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 )

What's this?

You are currently reading Interesting Read: Tear Down that GIL! at jessenoller.com.

meta