Merge extension with extended type

[pcantrell]

Continuing #289 based off of master instead of swift-2.0.

[pcantrell]

I went ahead and got this one ready, now that @jpsim big Obj-C push is done. It’s now based off of master instead of swift-2.0, and the specs are up to date.

This is ready to merge if you’re happy with the output. As a reminder of issues from the older discussion:

• @jpsim was concerned about clobbering // MARK: text with the “Default Implementation” heading. That’s now fixed.
• There was concern about Members added to extension of nested type show up in the parent #333, which turned out not to be related to this PR.
• Sourcekitten does not return extension declarations #334 limits the usefulness of this feature somewhat, but again, that problem existed before this PR.

[jpsim]

it seems nicer/cleaner to just do a string comparison if we're matching for an exact string anyway?

[jpsim]

This is so so great. I'm still impressed at how concise you managed to keep this code.

👍 other than my very minor comments.

Also, although this is something I haven't done yet, there are lots of areas now where Swift processing differs from Objective-C processing, and we should at some point go through the codepaths to branch off where needed to keep things efficient.

[pcantrell]

This is so so great. I'm still impressed at how concise you managed to keep this code.

That’s kind of you. I actually found it a little hard to read on re-review, so I did a very light refactoring that I think helps.

@jpsim, I’ve made your very sensible fixes, so if CI likes it, we’re good to go.

[jpsim]

Looks like there are rubocop violations to address:

Offenses:
lib/jazzy/sourcekitten.rb:313:9: C: Align the operands of an expression in an assignment spanning multiple lines.
        .group_by { |d| deduplication_key(d) }
        ^^^^^^^^^
lib/jazzy/sourcekitten.rb:314:9: C: Align the operands of an expression in an assignment spanning multiple lines.
        .values
        ^^^^^^^
22 files inspected, 2 offenses detected
RuboCop failed!
The command "bundle exec rake spec" exited with 1.

[pcantrell]

Yeah, weird — why didn’t those show up locally? Fixing…

[pcantrell]

Not getting that rubocop failure locally. I notice that I have rubocop 0.26.1 locally, which is what’s in the Gemfile.lock, but Travis is downloading 0.34.2. ?!?

Just pushed a commit that’s a shot in the dark at what I think rubocop is asking for. We’ll see if it passes.

[jpsim]

Rubocop is pinned at 0.34.2 in the Gemfile.lock on master: https://github.com/realm/jazzy/blob/master/Gemfile.lock#L88

In any case, your shot in the dark passed! 🎉

I'll add a changelog entry for you in the spirit of getting this in.

[pcantrell]

GTG

[pcantrell]

Oh, right , changelog. Thanks.