Hacker News new | ask | show | jobs
by Monkeyget 4283 days ago
Great project!

I've done some python code optimization recently and I was fairly disappointed in the profiling options; more precisely the profiling visualization options. Standard python provides pstats, which is basic and has an awkward command based GUI. I've tried runsnakerun and snakevis which do provide a nice graphical view but little else.

Easy to use, hierarchical list, clear output. I like this. I never used live profiling before and I can already see where it could be super handy.

Remarks:

-how do I pass arguments to my program. profiling seems to absorb all parameters

-Is it possible to access standard input/output when live profiling?

-Is it possible to reset the profiling statistics when live profiling? Eg.: I profile a webserver, profile a request, reset stats, profile another request,...

-I'd love to see an integration with IPython. Interactive optimization could be great: I profile some code, tweak it, profile again,...

2 comments

For IPython-based profiling, I find the %timeit and %memit magic commands invaluable. This post has a good overview: http://pynash.org/2013/03/06/timing-and-profiling.html
There is lsprofcalltree which will use normal cProfile and give you output suitable for kcachegrind (which is awesome)