Fix Sass deprecation warning re: reverse-gradient function.

[ry5n]

Since Sass 3.4.10 (January), passing anything other than a non-string value to the unquote function is deprecated. This affects the reverse-gradient function, which tries to pass a list through unquote. This PR removes this usage of unquote.

Status: Ready to merge

Reviewers: @jeffkamo @avelinet @cole-sanderson @mlworks @yourpalsonja @nastiatikk
Ticket: N/A
Linked PRs: N/A

Changes

• No longer attempt to unquote the final list returned from reverse-gradient.

How to test-drive this PR

• Clone this repo and run npm install && bower install to ensure dependencies are installed
• Run grunt to compile Sass
• Ensure all the DEBUG statements in output pass
• Open the test/index.html file and verify that the reverse-gradient function still works.

Research resources

• Sass commit adding unquote deprecation warning

[nastiatikk]

I didn't know about unquote thing

[nastiatikk]

So to clarify since Sass 3.4.10 you can pass non-string value only?

[ry5n]

@nastiatikk Yes, in 3.4.10+ you can’t pass a list to unquote, which we are accidentally doing in the reverse-gradient function. Attempting to unquote a list never did anything, so we can safely remove the call in this case. I’m actually not sure when unquoting is strictly necessary, if ever.

[nastiatikk]

Maybe @jeffkamo could give some ideas? Seems like it was his tweak 70d0b7b

UPD: But i see it was done by Kyle P originally

[jeffkamo]

I'm fine with this change. If the usecase is no longer supported (via deprecation) it's probably a good idea that we fix our code to reflect that.

[jeffkamo]

I just haven't tested the change yet.

[ry5n]

@jeffkamo Just to be clear, there was never a “use case” for passing a Sass list to unquote – it just allowed you to do that and passed the list through unchanged. Now, this is deprecated (and will be an error in future versions).

[jeffkamo]

Yup, that's basically what I meant. I suppose I should've said "non-use case" instead.

[ry5n]

🙇

[nastiatikk]

This change sounds reasonable
+1

[kpeatt]

I'm sure I must have done that for a reason though... 🐙

[ry5n]

@kpeatt Hi there 😄. I can see it being done as insurance, in case the list itself could be a string. Re-reading the code again, it seems like for some code paths that could be the case. We do need to change something here, but probably need to be more careful about it, and have a test for it. Let me look at this again and make sure.

[ry5n]

Opened #47 until I can write a more careful PR.