- Use `sort_by` to sort the array, rather than comparing each element
to the next.
- This doesn't error with complaints about clobbering at all when run on
`homebrew/cask`, hurray. And it also handles interpolations correctly,
rather than ignoring them.
Co-authored-by: Bevan Kay <email@bevankay.me>
- Interpolating the version into a path is a common pattern, but the interpolations
trip up the alphabetization autocorrect quite spectacularly, so let's
ignore them (for now?).
Without this, there's an infinite loop on `brew style --fix` if you have
a `if Hardware::CPU.arm?` in a `postflight` block where it will change
back and forward between `if` and `on_os` syntax forever.
- Part of issue 16323.
- Previously this was being done manually by Cask maintainers.
- While we're here, enforce that the `zap trash` path is not in `[]` if
it only contains a single element.
- This is buggy on actual Casks, hence the draft PR.
QuicTLS is a fork of OpenSSL that adds support for QUIC. We'll probably
end up adding it to homebrew/core at some point (see
Homebrew/homebrew-core#134975), but I don't think we want to actually
use it as a dependency of any formulae in place of OpenSSL.
We ought to only allow it for software that actually require QuicTLS in
place of OpenSSL, but I'm not aware of any existing formulae that have
this requirement.
After some discussion, we decided to change the DSL to get rid of
the `plist_name` and `service_name` methods which aren't meaningful
for most users.
The new DSL looks like this:
```rb
service do
name macos: "name", linux: "name"
end
```
I also updated some specs here to reflect these changes.
There was some talk about maybe deprecating `plist_name` and `service_name`
but I think that's outside of the scope of this PR so I'm leaving
them as is for now. One benefit of this is that everything here is backwards
compatible.
It now checks for two new edge cases.
1. Are the required methods defined in the block?
2. Is the `run` command defined if we're using methods other
than `service_name` or `plist_name`?
It has also been updated to only get the list of method call nodes
once instead of doing it repeatedly for each cellar path check.
- Thanks to Markus on Slack for saying "the cop should only apply to the
content of the blocks, or more specifically only to stanzas that are
direct children of cask or on_* blocks", which made me realize that
I was overcomplicating things.
- Now that we detect correct stanza _grouping_ within `on_*` blocks in
Casks (PR 15211), correct stanza _ordering_ in `on_*` blocks was the
next logical step. For example, `url` has to come after `version` and
`sha256` in an `on_macos` or `on_intel` block for consistency with the
top-level stanza order we enforce elsewhere.
- Still not doing the nested `on_os` inside `on_arch`, that felt
excessive for an edge case that isn't present in any actual real
Casks we have. I removed the test with that specific TODO.
- Since moving `comments_hash` to `Stanza`, we've been using the wrong
kind of "comments": the comments for the _stanza_, not the comments
for the entire Cask.
- Add a test to ensure this actually works. There was previously an
infinite loop here due to the bad `comments`, visible in a `StanzaOrder`
cop test, which I speculatively added a failing test for. Turns out
that supporting nested stanza _ordering_ (vs. just grouping) is a
whole separate piece of work (there are multiple TODOs there already),
so I've backed that out and will do that separately.
