Add a more meaningful "ptp"

[ebressert]

The peak-to-peak function, ptp(), is a simple and handy function to use, but it confuses the PEP20 suggestions a bit. See this discussion for reference: http://d.pr/i/snXG

Based on the discussion, it may be a good idea to make the name more relevant? Some suggestions include span() or valuerange().

[adrn]

+1 to this idea

[andycasey]

Also +1

[rgommers]

Makes sense, I like valuerange. ptp should be kept as an alias though, it's not necessary to break existing code for this.

Could you propose this on the mailing list? There's just a discussion going on about some other functions to add for 1.8.

[ebressert]

@rgommers Sure. I'll send something out in today regarding this.

[charris]

Still open. Was there a discussion?

[WarrenWeckesser]

FWIW: Over in ufunclab, I created a function called peaktopeak as an exercise in creating generalized ufuncs. peaktopeak makes two imrovements over ptp:

• It makes a single pass over the data, so it will usually be faster than ptp.
• For arrays of signed integers, peaktopeak returns an unsigned integer result. ptp returns a negative value with arrays of signed integers when the peak-to-peak value exceeds the maximum signed integer of the array's dtype (e.g. np.array([-100, 0, 100], dtype=np.int8).ptp() returns -56).

[mattip]

Closing. This is more appropriate for scipy.