Fixed an issue with lengths being undefined.

[barcellano-envoy]

Also, it looks like this was added at some point to the compiled lib and then removed.

[aabenoja]

Added that comment to the wrong spot. My apologies.

[barcellano-envoy]

I believe the issue is in @keyframes, eg

@-webkit-keyframes sound-wave {
  0% {
    opacity: 0.6;
    -webkit-transform: scale3d(0, 0, 0);
    -moz-transform: scale3d(0, 0, 0);
    -ms-transform: scale3d(0, 0, 0);
    -o-transform: scale3d(0, 0, 0);
    transform: scale3d(0, 0, 0);
  }

  100% {
    opacity: 0;
    -webkit-transform: scale3d(1, 1, 1);
    -moz-transform: scale3d(1, 1, 1);
    -ms-transform: scale3d(1, 1, 1);
    -o-transform: scale3d(1, 1, 1);
    transform: scale3d(1, 1, 1);
  }
}

How do I add a test and run the tests?

[aabenoja]

Look at the test folder and you'll find two folders. The input folder contains css files that may or may not exceed the selector limit. The output folder contains folders corresponding to a matching input css filename. The output css files are 0.css, 1.css, 2.css depending on the length of the input css file for that test.

For these keyframes test I suggest modeling after the media queries test.

[barcellano-envoy]

I did create the css file for the test. How do I run the tests?

[aabenoja]

In a local clone with the changes I suggested I did not run into an undefined nestedRule.selectors, mainly because the keyframes at-rule does not contain any selectors. I tried an empty media query but the css parser module was smart enough to give nestedRule.selectors an empty array.

As far as testing is concerned run npm test in your console.

[barcellano-envoy]

Forgive my coffeescript, never really used it much. I have updated the code with your suggestion, and also added a test for the keyframe case

[aabenoja]

@paulyoung Got any input?

[barcellano-envoy]

Ok, I also figured out another test case where even this fails

 @media screen and (max-width: 1199px) {
    #selector {
        color: #fffff;
    }
    @media screen and (max-width: 991px) {
        #selector {
            color: #fffff;
        }
        @media screen and (max-width: 767px) {
            #selector {
                color: #fffff;
            }
        }
    }
}

where

for nestedRule in rule.rules
          numNestedRuleSelectors += nestedRule.selectors.length

nestedRule.selectors can be undefined.

What's the best coffeescript way to resolve this?

[aabenoja]

As much as I dislike nested media queries the current implementation will NOT hit that third media query. This is most likely also breaks if a rule contains a media query as well. Thanks for catching this ugly excuse for valid css.

[barcellano-envoy]

It's what libsass compiles the SASS we have to. :( Ruby sass does not.

[barcellano-envoy]

What do you think is the best solution? Recursive function to travel down the tree?

[aabenoja]

I believe a recursive function would be best in this case, yes. We'll need to apply it to both rules and media queries.

[barcellano-envoy]

I'm pretty naive on my coffeescript skills. Want me to give it a go, or would it be better to wait for someone closer to the project?

[barcellano-envoy]

Actually, according to the test suite, this makes the test pass

for nestedRule in rule.rules
          break if typeof nestedRule.selectors is 'undefined'

          numNestedRuleSelectors += nestedRule.selectors.length

Will commit for you to see

[barcellano-envoy]

Ahh, I see. It probably wouldn't break the file on the correct selector

[aabenoja]

I say give it a go. I'm not very coffeescript savvy myself; I personally prefer to write my javascript by hand. I have something in mind. I'll submit a PR to your fork in a bit.

[barcellano-envoy]

Ok, I'm going to write a few more tests.

[aabenoja]

Well, the css parser is blowing up trying to break apart nested media queries. Good to know.

[aabenoja]

So the css parser used in this project is blowing up trying to parse those nested media queries. It dies completely every time it tries on my end, at least. I suppose now is as good a time as ever to try PostCSS.

[barcellano-envoy]

The test I wrote passes. Not sure why, but it does. I'm sure it's a flaw with the test probably.

[aabenoja]

npm run build then npm test and tell me if it doesn't break.

Scratch that, I was still running off of modified node modules.

[aabenoja]

I must have had a bad character somewhere in my test file that caused the blowing up. Weird, but fine.

[aabenoja]

I'll be updating the unit test again shortly.

[paulyoung]

Wow, don't know how I missed this one for so long. I'll review when I have more time.

Thanks for the effort, and sorry for the silent treatment!

[aabenoja]

This is no longer valid with #63.