3 ms·
I like Python (and Ruby) a ton, but they may find themselves legacy-stamped if they don't develop a good concurrency story quite soon. No big deal to be ~50 ti
by slurry 14y ago
I like Python (and Ruby) a ton, but they may find themselves legacy-stamped if they don't develop a good concurrency story quite soon.
No big deal to be ~50 times slower than native code in a single-thread, single-processor use case, but for a well-parallelizable problem on a multicore machine you end up being several multiples of ~50 times slower.
Google just gave up on trying to optimize Python (see Unladen Swallow project) and told the engineers to move on to something useful. I am not sure if the interanls are just to messed up too make it work or if resistance and denial on the part of "B"DFLs and the community is too strong.
But when javascript - JAVASCRIPT! - has a better server-side concurrecy pitch than your language, it is in big trouble
- mitchi 14y agoInteresting story. Do you know they moved on to what? Java?
- slurry 14y agoPython is still used alongside C++ and Java (and some Go) but primarily for prototyping and small utility scripts. A Google engineer explains: "Well, simple common sense is going to limit Python's applicability when operating at Google's scale: it's not as fast as Java or C++, threading sucks, memory usage is higher, etc. One of the design constraints we face when designing any new system is, "what happens when the load goes up by 10x or 100x? What happens if the whole planet thinks your new service is awesome?" Any technology that makes satisfying that constraint harder -- and I think Python falls into this category -- should be discouraged if it doesn't have a very strong case made in its favor on other merits." And note this was before the failure of Unladen Swallow, before van Rossum's departure.