zlacker

Who is pro-GIL

Nobody is pro-GIL per se. But a lot of people were pro-No-Single-Threaded-Performance-Desegregation. The first GIL-removal patch was submitted all the way back against python 1.4, and regular attempts have been made ever since, but it wasn't possible to remove the GIL without making the single threaded performance of existing python code worse, so Guido and co. refused to accept them.

>>dagw+(OP)
And the performance hit is still up to 50%, just like in the first attempts. Except this time it comes from Facebook.

>>bgwalt+R1
Poking around, I've not been able to find anyone else claiming higher than 10%? Is this 50% from your benchmarks, or is there a link I can see?