Add short-circuit return to matplotlib.artist.setp if input is length 0

[samsontmr]

Fixes #7784

Before this patch, an IndexError is raised in matplotlib.artist.setp if argument is an empty list,

import matplotlib
matplotlib.artist.setp([])

[anntzer]

Just check if not objs: instead (objs is a list defined immediately below).

[samsontmr]

Wouldn't if not objs return false if objs is a list of empty lists? Or do you mean to check for if not obj?

[tacaswell]

Agree using if not objs: would be better. I would also move this down to L1453 (right above the line with objs[0]) after we have normalized the input to a list.

[samsontmr]

Oh I see! The if else statements guarantee a list, right?

[tacaswell]

Yes.

[samsontmr]

Commits squashed

[tacaswell]

Looks good! Could you also add a test to test_artist.py that exercises this code path?

[phobson]

Apologies if I'm missing something obvious here, but couldn't this be at the very top of the function? e.g.,

    if not objs:
        return
    elif not cbook.iterable(obj):
        objs = [obj]
    else:
        objs = list(cbook.flatten(obj))

[tacaswell]

@phobson users will find pathological cases 😉

In [62]: bool([[]])
Out[62]: 
True

[phobson]

@tacaswell wow. great catch with that.

[samsontmr]

@phobson objs doesn't exist at the top either, only obj ;)
@tacaswell will do!

[codecov-io]

Current coverage is 62.10% (diff: 100%)

Merging #7801 into master will decrease coverage by 0.01%

@@             master      #7801   diff @@
==========================================
  Files           174        174          
  Lines         56028      56052    +24   
  Methods           0          0          
  Messages          0          0          
  Branches          0          0          
==========================================
+ Hits          34803      34809     +6   
- Misses        21225      21243    +18   
  Partials          0          0          

Powered by Codecov. Last update 9b1ece0...193da2e

[samsontmr]

@tacaswell Tests added

[NelleV]

LGTM

[NelleV]

I've edited your comment to explain what it fixes. It saves us reviewers a lot of time not to have to identify what the original problem was.

[tacaswell]

This should be backported after 2.0

@samsontmr Thank!

[QuLogic]

This doesn't backport that easily, or at least, I'm not familiar enough with that section of the code to do it.

[tacaswell]

@QuLogic Lets drop the backport on this.

[QuLogic]

Works for me.