eliminate normalize_slice in favor of a standard Python idiom

[jasongrout]

Also, it would be good to optimize it if possible.

[04:21] <craigcitro> yeah, file a ticket for that and assign it to me.
[04:21] <jason-> (couldn't get cimport to work...)
[04:22] <craigcitro> well, the pari gen.pyx probably wasn't the best place for that function
[04:22] <craigcitro> so it needs to be moved anyway

CC: @jasongrout

Component: misc

Issue created by migration from https://trac.sagemath.org/ticket/4974

[craigcitro]

comment:1

patch not necessarily ready for review yet.

[craigcitro]

comment:2

I think this is good to go.

[sagetrac-ylchapuy]

comment:3

Some questions/remarks.
Shouldn't the following always be true ? (at least this is what I understand from the docstring)

normalize_slice( s , size_list) == range(size_list)[s.start:s.stop:s.step]

If no (which is the case actually) this looks quite incoherent to me;
if yes, is there a reason not to rewrite this fonction as a one liner?

[sagetrac-ylchapuy]

comment:4

forget the one liner thing it's obviously a bad idea

[craigcitro]

comment:5

Wait, so disregarding whether or not we'd want to implement things that way -- are you still saying there are cases where that isn't true?

[jasongrout]

comment:6

There is an inconsistency with standard python:

[12:12] <jason-> sage.misc.misc_c.normalize_slice(slice(2,None,-1),3)
[12:12] <jason-> gives me 2
[12:12] <jason-> but 
[12:12] <jason-> sage: range(3)[2::-1]
[12:12] <jason-> [2, 1, 0]

[craigcitro]

comment:7

Jason Grout just brought up an example in IRC where exactly that would happen. Here's a case:

sage: range(3)[2::-1]
[2, 1, 0]

normalize_slice will not return that, since it interprets a missing stop to be equal to size. This isn't documented as part of the Python semantics, but I think we should switch it to behave the same way regardless. (Also, it'll be easy to implement.)

This is now needs work until I change this.

[craigcitro]

comment:8

Another case brought up by Jason Grout:

sage: range(5)[-6:]
[0, 1, 2, 3, 4]