TST: correct constructor in extension array tests

[jorisvandenbossche]

Replace type(data)(..) with data._constructor_from_sequence.

A bit annoying the name is that long :-), but this is the correct usage.

cc @TomAugspurger

[codecov]

Codecov Report

Merging #20746 into master will decrease coverage by <.01%.
The diff coverage is n/a.

@@            Coverage Diff             @@
##           master   #20746      +/-   ##
==========================================
- Coverage   91.85%   91.84%   -0.01%     
==========================================
  Files         153      153              
  Lines       49308    49303       -5     
==========================================
- Hits        45290    45284       -6     
- Misses       4018     4019       +1

Flag	Coverage Δ
#multiple	90.23% <ø> (-0.01%)	⬇️
#single	41.89% <ø> (ø)	⬆️

Impacted Files	Coverage Δ
pandas/core/internals.py	95.53% <0%> (-0.05%)	⬇️
pandas/core/frame.py	97.16% <0%> (-0.01%)	⬇️
pandas/core/arrays/base.py	84.14% <0%> (ø)	⬆️
pandas/core/resample.py	96.43% <0%> (ø)	⬆️
pandas/core/indexes/datetimelike.py	96.72% <0%> (ø)	⬆️
pandas/core/indexing.py	93.08% <0%> (ø)	⬆️
pandas/core/generic.py	95.94% <0%> (ø)	⬆️
pandas/tseries/offsets.py	97% <0%> (ø)	⬆️

---

Continue to review full report at Codecov.

Legend - Click here to learn more
Δ = absolute <relative> (impact), ø = not affected, ? = missing data
Powered by Codecov. Last update 0ae7e90...88559d7. Read the comment docs.

[TomAugspurger]

Thanks.

I've grown to dislike the name _constructor_from_sequence even more in the interim. It sounds like the method is returning constructor, which would then be used to construct the array. Instead it's creating a new array from a sequence of scalars (so _from_scalars...)

Is it worth changing before it's too late?

[jreback]

why dont we rename to _from_sequence?

[jorisvandenbossche]

I think we can certainly leave out the constructor part. For _from_scalars or _from_sequence, the doesn't indicate it is from a list/array of scalars, and the second does not indicate that it is from the dtype's scalar objects.
I personally have a preference for _from_scalars because I find the 'scalar' the more important information of the two, but _from_sequence is certainly already an improvement compared to the current name.

[TomAugspurger]

Either of those are fine. Slight preference for `_from_scalars`, for the reasons Joris outlined. But from the context, I think it's clear enough what `_from_sequence` is doing.

…

On Fri, Apr 20, 2018 at 2:13 AM, Joris Van den Bossche < ***@***.***> wrote: I think we can certainly leave out the constructor part. For _from_scalars or _from_sequence, the doesn't indicate it is from a list/array of scalars, and the second does not indicate that it is from the dtype's scalar objects. I personally have a preference for _from_scalars because I find the 'scalar' the more important information of the two, but _from_sequence is certainly already an improvement compared to the current name. — You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub <#20746 (comment)>, or mute the thread <https://github.com/notifications/unsubscribe-auth/ABQHIs7TwlIFipdK_GOKnyUkwkcxIWCrks5tqYqpgaJpZM4Tb6xW> .

[jreback]

let's do the rename to _from_sequence in this PR (and update docs etc)

[TomAugspurger]

The Appveyor failure was an HTTP issue. @jorisvandenbossche you can merge whenever.