remove heavy dependencies about pa11y HOT 5 CLOSED

From a quick scan, it looks like the following functions are used from the _ module:

• _.each
• _.extend
• _.clone
• _.countBy
• _.defaults
• _.groupBy
• _.pick

That's a fair bit of functionality that would need to be replicated by another module?

from pa11y.

not replicated by another module, but other modules. this is just my personal ocd, but smaller modules tend to perform better.

from pa11y.

Something to consider, if you have any suggestions for replacement modules @stephenmathieson? I'm not going to focus heavily on this kind of optimisation (if needed at all) until the adding of new features has settled down a bit – from past experience, trying to optimise while everything's changing under your feet is a painful task.

I'll pop this on the back-burner for now, and will revisit after the next release.

from pa11y.

this is definitely not a big deal, and certainly shouldn't block other more important changes from making their way into the codebase. i agree that if this sort of path is to be followed, it can wait.

regarding replacements:

• _.each = [].forEach or for (var prop in obj) fn(prop, obj[prop]) or Object.keys(obj).forEach
• _.extend = utils.extend or for (var prop in obj) new_obj[prop] = obj[prop]
• _.clone = [].slice.call or utils.extend({}, obj) or clone
• _.defaults = _.extend = utils.extend = ...

for methods like _.pick and _.groupBy, i'd have to suggest a slight refactor (or at least a deeper examination as to why these methods are needed)

from pa11y.

Reading over this again, I might revisit the dependencies at some point but underscore is working fine for now. If I was to move away from underscore, it would be towards several small modules which replace the different functions being used. I wouldn't like to reimplement the functionality in this repo as it would increase the complexity and amount of code we'd have to test/maintain.

from pa11y.