Improve fancy indexing when selecting cross-sections (original) (raw)

Something like

df.ix[0, ['C', 'B', A']]

is currently performing a wasteful reindex(columns=['C', 'B', A']), need to have some better way to infer the "right" order to perform the operations in