Java, XQuartz and the CLT separate header package aren't required for
everyone's Homebrew usage or a default macOS development install.
As a result, only show then in `brew config` when they are actually
installed.
Use the environment variables set by `brew test-bot`. Eventually we'll
disable Travis CI running CodeCov so move `TRAVIS` references to
`HOMEBREW_TRAVIS_CI` so it doesn't need whitelisted.
Also, fix `azure-pipelines.yml` so it's testing the correct version of
Homebrew/brew (the one checked out in the `pwd`).
Adjust the rules based on the current codebase. Remove various enable,
disables and default values that are unnecessary. Add more comments
explaining why. Make minor changes needed to enable a few more rules.
Use 124 max line length everywhere. Also, reduce tap max line length to
189 as Homebrew/homebrew-core has that as a maximum now. In future
Homebrew/homebrew-core will also be reduced to 124 maximum line length.
Rather than relying on a `HOMEBREW_FORCE_BOTTLE` variable (which ends
up doing silly things like forcing bottle usage even when options are
provided) instead handle this at the `or_later` bottle detection
level so on prerelease versions of macOS any bottle looks like an
`or_later` bottle (unless various environment variables are set).
Fixes issues noted in:
https://github.com/Homebrew/brew/pull/4520#issuecomment-407229605
1. Running `brew linkage some_package` does not set the cache.
2. Running `brew linkage --cached some_package` when `DatabaseCache.empty?` returns `true` should build the cache.
3. Running `brew linkage --cached some_package` when `DatabaseCache.empty?` returns `false` should use the cache.
Ensure that a non-zero exit code is set both for missing random dylibs
and random missing dependencies.
Additionally, while we are here, drastically trim down the public
interface for this class to the bare minimum and allow getting the
output from `display_test_output` as a variable.
Fixes issue mentioned by @ilovezfs in:
https://github.com/Homebrew/brew/pull/3940#issuecomment-383794520
The `cask` attribute doesn't make as much sense on Linux but can be
ignored there. The advantage of this change is that (like #4086) it
allows figuring out the relevant cask for a formulae requirement on a
Linux machine.
