Fix objects with functions - replace (objToCss) with (objToCssArray)

[d7my11]

Fixes: #2468
Addresses: #2414

---

The issue was in objToCss method that didn't handle methods for objects.

• Replace objToCss with objToArray which handles methods as well.
• Cover createGlobalStyle and styled with more test cases.

[mrmckeb]

This is great work, thanks! Out of interest, are there any performance metrics around a) objects/functions over template literals, and b) methods as properties?

[d7my11]

@mrmckeb I'm not 100% sure actually, but here is my understanding:
using template literal will be converted by babel to arrays if they have embedded expressions and get .concat()ed at the end.

Using objects gets converted by styled-components and it gets slower with nested objects because you have to go through every key for each object.

So, based on my assumptions, using template literal is faster specially if the object contains nested objects.

I would like to hear from somebody who is fully aware of the project to verify or correct me if i'm wrong.

[mrmckeb]

Good information, thanks @d7my11 - and great work.

[quantizor]

I'd also love if you could do this against canary instead or open a second PR for it.

[d7my11]

@probablyup are there any minor releases for bugs before merging canary into master? if so, will open a second PR for this. If not, will change the base branch.

[quantizor]

Yeah 4.x isn’t closed out yet

…

On Thu, May 2, 2019 at 3:29 AM Abdel-Rhman ***@***.***> wrote: @probablyup <https://github.com/probablyup> is there any minor releases for bugs before merging canary into master? if so, will open a second PR for this. If not, will change the base branch. — You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub <#2489 (comment)>, or mute the thread <https://github.com/notifications/unsubscribe-auth/AAELFVVLNTA4SWXAIFAMUM3PTKQ6XANCNFSM4HEWR53Q> .

[d7my11]

Will open another PR then

[quantizor]

If you could build it on top of @kitten's latest PR that'd be great as I think he made some changes to code near what you're changing

[d7my11]

@probablyup opened another PR for canary #2542

[quantizor]

@d7my11 can you get this rebased against master? Looks like a test file just needs a little fix

[d7my11]

@probablyup done

[quantizor]

this would be cleaner as a reduce I think

[d7my11]

true, but it's slower than .forEach because the result has a different type than the items. i did a quick benchmark for that here

[quantizor]

Fair enough!

[quantizor]

Last thing, a changelog entry would be lovely

[quantizor]

LGTM and thanks!

[d7my11]

@probablyup thanks for reviewing