Move Cask core code to Homebrew

[AnastasiaSulyagina]

• Have you followed the guidelines in our Contributing document?
• Have you checked to ensure there aren't other open Pull Requests for the same change?
• Have you added an explanation of what your changes do and why you'd like us to include them?
• Have you successfully run brew tests with your changes locally?

---

Moving Cask core code to Homebrew as part of #14384.

Change is splitted in 2 commits to be easier to review. Any advice on better splitting or anything else welcome.

cc @MikeMcQuaid @vitorgalvao @jawshooah @adidalal (bring other cask maintainers please)

[jawshooah]

Pinging @reitermarkus, @mwean, @fanquake, @victorpopkov, @alebcay, @claui, and @phinze for visibility.

[MikeMcQuaid]

CC @Homebrew/maintainers as this is a big one.

[MikeMcQuaid]

I've opened #727 to fix the test failures here as they are in our code making assumptions about Cask code that we need to decide if they are valid.

[MikeMcQuaid]

Some of the cask-tests are failing: https://bot.brew.sh/job/Brew%20Pull%20Requests/1302/version=el_capitan/testReport/junit/brew-test-bot/el_capitan/cask_tests/

[MikeMcQuaid]

You can probably remove all the "under Ruby 2.1" tests and may need some test fixtures to not assume Caskroom/homebrew-cask is tapped (although we could compromise that and just tap it before the test). It's probably worth doing these in a separate PR before merging this or the respective Cask PR.

[AnastasiaSulyagina]

Seems that cask tap's path for test-bot is not HOMEBREW_LIBRARY.join("Taps", "caskroom", "homebrew-cask"). Will take a look

[MikeMcQuaid]

Seems that cask tap's path for test-bot is not HOMEBREW_LIBRARY.join("Taps", "caskroom", "homebrew-cask"). Will take a look

@AnastasiaSulyagina on the test-bot: it's not even tapped 😀. This is something we could address at a later point, though.

[MikeMcQuaid]

This is something we could address at a later point, though.

Although let me know which you'd rather do.

[AnastasiaSulyagina]

I'd rather tap it before tests.
Why to address it at later point and not now?

[MikeMcQuaid]

I'd rather tap it before tests.

🆒 I can open a PR for that.

Why to address it at later point and not now?

Basically because at some point we're going to need to remove that reliance on having the tap tapped in favour of making tests create the Casks they expect and I figure that doesn't need to necessarily block this PR being merged.

[AnastasiaSulyagina]

I can open a PR for that.

That would be nice :)

Basically because at some point we're going to need to remove that reliance on having the tap tapped in favour of making tests create the Casks they expect and I figure that doesn't need to necessarily block this PR being merged.

Thanks, got it.

[reitermarkus]

Speaking of Taps, If possible, you might want to merge in the last two commits in Homebrew/homebrew-cask#23821. The only thing I'm missing is an elegant way to redefine Tap::TAP_DIRECTORY for our tests.

[MikeMcQuaid]

@reitermarkus I think you could merge that in for now as there's already conflicts there until there's a 👍 on holding off merging core PRs.

[reitermarkus]

Ok, merged the three remaining core PRs. Holding off now until this move is complete.

[DomT4]

Do we need stuff like the GitHub templates, CoC & other dotfiles from Cask here? Are they not remaining with the Cask tap?

[MikeMcQuaid]

Do we need stuff like the GitHub templates, CoC & other dotfiles from Cask here? Are they not remaining with the Cask tap?

@DomT4 I'm in favour of moving basically everything for now and then removing them once they are over; it's too hard to review the diff for now.

[DomT4]

Cool. That was exactly my thought, that removing the unnecessary text files might make the diff actually reviewable, heh, but whether that's done in pre or post doesn't matter much.

[AnastasiaSulyagina]

@jawshooah, are these tests needed? If yes, why should directories be undeletable and should I edit the paths somehow to fix it?

rspec ./spec/cask/macos_spec.rb:43 # OS::Mac says the user library directory is undeletable
rspec ./spec/cask/macos_spec.rb:31 # OS::Mac says the home directory is undeletable

[jawshooah]

Undeletable directories include system and user directories that should never be included in uninstall :delete or zap :delete. Those tests ensure that, were a Cask author to mistakenly include, for example, ~/Library in uninstall :delete, we would not actually delete that directory.

[AnastasiaSulyagina]

As I understood /Users/brew/Jenkins/pr-brew/version/el_capitan/home/ is Dir.home for test-bot. In macos.rb there is ~ string which actually should equal Dir.home. Am I wrong?

[reitermarkus]

This would mean we have need to call sub(%r{^~(?=(/|$))}, Dir.home) everywhere we expect to receive a relative path instead of (or in addition to) expand_path.

[AnastasiaSulyagina]

Hm. I thought that as it never was an issue before and appeared now in tests only, I can just replace ~ with Dir.home in Undeletables list. If it's unacceptable I'll fix it

[AnastasiaSulyagina]

Why can build pass in travis-ci/pr and fail in default?

[MikeMcQuaid]

@AnastasiaSulyagina The filesystems may have different permissions.

[reitermarkus]

@AnastasiaSulyagina, could you rather cange https://github.com/AnastasiaSulyagina/brew/blob/move-cask/Library/Homebrew/cask/lib/hbc/macos.rb#L364 to .map { |x| Pathname(x.sub(%r{^~(?=(/|$))}, Dir.home)).expand_path } instead of changing the whole list, to keep it at little more readable?

[MikeMcQuaid]

Mentioned to @AnastasiaSulyagina in Slack:

brew/Library/Homebrew/dev-cmd/test-bot.rb

Lines 942 to 943 in b39eba6

	ENV["HOMEBREW_HOME"] = ENV["HOME"] = "#{Dir.pwd}/home"
	mkdir_p ENV["HOME"]

is the issue; the test-bot overrides the $HOME variable to stop things being dumped in ~ and I'm guessing it's not getting expanded consistently as a result.

[reitermarkus]

@MikeMcQuaid, might be worth knowing what File.expand_path is using internally then to see if we can also override it globally in tests.

[reitermarkus]

Ok, seem's like it is using $HOME internally as well: https://github.com/ruby/ruby/blob/7a4aea47e281402a178b69b5f90d6ed330122e67/file.c#L3209

[AnastasiaSulyagina]

Rebased both PRs, fixed macos.rb

[reitermarkus]

I might have found the reason why some of the undeletable? tests are still failing.

irb> File.identical?('does-not-exist', 'does-not-exist')
=> false
irb> File.identical?('exists', 'exists')
=> true

Seems like File.identical? doesn't work on nonexistent paths.

@AnastasiaSulyagina, could you try changing https://github.com/AnastasiaSulyagina/brew/blob/move-cask/Library/Homebrew/cask/lib/hbc/macos.rb#L369-L375 to the following?

  def system_dir?(dir)
    SYSTEM_DIRS.include?(Pathname.new(dir).expand_path)
  end

  def undeletable?(dir)
    UNDELETABLE_DIRS.include?(Pathname.new(dir).expand_path)
  end

[AnastasiaSulyagina]

1) Hbc::SystemCommand with a very long STDERR output returns without deadlocking
     Failure/Error: wait(10).for {
       RSpec::Wait::TimeoutError
     # ./spec/cask/system_command_spec.rb:136:in `block (3 levels) in <top (required)>'

It is not related to the PR, is it?

[reitermarkus]

@AnastasiaSulyagina, could you bump https://github.com/AnastasiaSulyagina/brew/blob/move-cask/Library/Homebrew/cask/spec/cask/system_command_spec.rb#L136 to 15 seconds? I don't think there's actually a deadlock, as this only times out occasionally.

[reitermarkus]

Finally! 🎉

[MikeMcQuaid]

Merging this and Homebrew/homebrew-cask#23852 should follow soon afterwards 🎉. Great work, @AnastasiaSulyagina!