ENH: print more detailed truncated sequence if a limit is given

[cpcloud]

works similar to numpy except that the edgeitems is overriden by threshold rather than the other way around which is what np does.

closes #3391

• add release notes

[cpcloud]

@y-p can u take a look-see please

[cpcloud]

AH HA! i figured out the case where the overlap happens now i can write a test :) thanks @y-p for pointing that out

[cpcloud]

@y-p

numpy's edgeitems (a new option i'm introducing called max_edge_items) overrides threshold (equivalent to our max_seq_items), when threshold < edgeitems.

i don't think numpy's behavior in this regard is necessarily what a user would expect. personally, i would expect max_seq_items to be a "hard limit" on the number of items that repr including the items repr'd on the edge of the sequence, thus i think max_seq_items should override max_edge_items. the edgeitems would then be forced to be at most n / 2 of the sequence, guaranteeing no overlap.

what do you think? this, of course, will not be consistent with numpy. right now the above described behavior is what i'm doing, but i can change if consistency with numpy is preferred.

[cpcloud]

here's an example from the tests i've added:

with cf.option_context('display.max_seq_items', 2, 'display.max_edge_items', 5):
    s = 'Int64Index([0, ..., 9], dtype=int64)'
    res = repr(tm.makeIntIndex(10))
    assert_equal(s, res)

[ghost]

I agree. when I suggested aligning with numpy I meant the format for summary: "foo foo foo ... foo foo foo", i.e.
symmetric. what you propose sounds right to me.

As simple as possible but no simpler.

[cpcloud]

for the special case of repring a MultiIndex when max_seq_items == 1 i think the repr should be

MultiIndex
[(u'C_l0_g0', ...), ...]

[ghost]

Put constraints on the setting of of max_seq_items (even number >=2), and save yourself the special cases.
Yes, for MultiIndex, considering each tuple as an item makes sense.

[cpcloud]

that certainly simplifies my life a lot

[cpcloud]

note to self remove this

[cpcloud]

@y-p can u take a look when u get a chance? thanks

[jreback]

@cpcloud close/push to 0.14?

[jreback]

@cpcloud ?

[jreback]

@cpcloud ?

[cpcloud]

pushing to 0.14

[jreback]

@y-p must have been accident

[ghost]

It's stale, but it needs to happen for 0.14. If @cpcloud doesn't finish, I'll do it.

[jreback]

@y-p gr8...i think i got all the accid closed ones...weird

[jreback]

@cpcloud what's the state of this?

[cpcloud]

I'll try to get this cleaned up this weekend.

[cpcloud]

Something happened just now when I tried to update the remote .... getting the infinity commits again.

[cpcloud]

Oh fudge I had those in a stash that has since been deleted .... ah well ... nothing major anyway. I'll open another PR.