cam macrae - blog

Smart guys say smart things

Many moons ago I read Bruce Eckel's Thinking In Java so his name rang a bell when I read this post on Python 3000. Although arguably a mess of confusion as regards language specification (Python 3000) and library (most of the post), it elicted a measured response from the BDFL, and bit of buzz around the ever-simmering Kill GIL sentiment in some sections of the Python community.

Aren't we done with this?

For the vast majority of problems I encounter, CPython is fast enough. Where it isn't I suck it up and write some C, or if it seems appropriate move that piece of the puzzle into the JVM (via a webservice, or whatever). I doubt I'm Robinson Crusoe here, but there is hardly a Python related newsgroup, mailing list or blog on which I lurk or contribute that doesn't do the GIL discussion at least once a month.

Round and round and round she goes; when the patches arrive we do not know!

Processes

There are those who think a threading model is the be-all and end-all of concurrent programming, those who couldn't care less, and those for whom threads are the satan's very own spawn (geddit?!?). No matter where you position yourself on the scale, you are bound to have been a member of the latter camp (or soon will be) at some stage.

I urge all thread junky / GIL haters (and everyone else) make a slightly-more-than-cursory study of erlang if you haven't already. There is a whole other world out there...

One of the more interesting things to come out of the discussion started by Bruce Eckel has been the many fingers pointing in the direction of pp. I'm yet to play with it, but pp (or something like it) might silence the GIL debate once and for all.