.ix performance regression?

[wesm]

is this kind of use cases reflected in our vbenchmarks? suppose @jreback is the one to ask here. more evidence need to rewrite the internals of pandas objects in C to reduce indexing overhead, among other things

http://stackoverflow.com/questions/17538902/is-there-a-way-to-speed-up-this-pandas-function

[jreback]

Current timings / I added to vbench

frame_get_value                              |   0.0017 |
frame_get_value_ix                           |   0.0033 |

OP should be using .get_value/.at or using a vectorized solution in any event

UPDATE:

these are the same on 0.10.1 and 0.12rc1

no regression on any of the indexing routings (to current master)

[jreback]

benchmark added: 7b2eaa4