This reduces the surface area of our `Kernel` monkeypatch and removes
the need to `include Kernel` in a bunch of modules.
While we're here, also move `Kernel#require?` to `Homebrew` and fully
scope the calls to it.
This should help further debugging.
Also, no need to formulae that aren't `*.rb` files as they'll be rejected
later anyway (and already were before we changed the
`Homebrew::EnvConfig.forbid_packages_from_paths?` defaults).
We need to handle symlinks in both directions for formulae in the
cellar/taps/cache otherwise we can get the nonsensical situation of
telling people a formula must me in a tap when it already is (but is
symlinked from elsewhere).
- Ensure that we're overriding it where necessary internally (but not
allowing this to be used as a public opt-out API)
- Better handle formulae from paths that look like taps
- Don't check if the `realpath` includes `/`: it always will.
Instead, see if the originally user-passed `ref` includes `/` or ends
with `.rb`.
- Always raise an error message, don't silently return.
- Allow cache paths in FromPathLoader when HOMEBREW_FORBID_PACKAGES_FROM_PATHS is set
- Fixes issue where HEAD installations fail due to temporary source downloads
- Add test case to verify cache paths are allowed when path restrictions are enabled
The issue occurred because HEAD installations download formula sources to cache
directories, but HOMEBREW_FORBID_PACKAGES_FROM_PATHS only allowed paths from
HOMEBREW_CELLAR and HOMEBREW_LIBRARY/Taps, causing the installation to fail.
Closes: homebrew/brew#issue-number
The `Formulary` and `system_command` debug output is incredibly verbose
by default and this is pretty annoying when all you want is to get
better backtraces when there's an error.
Instead, let's require `--verbose` and `--debug` for the noisiest output
message types.
Formulae can execute arbitrary Ruby code when being loaded. In
particular, they can print to stdout with methods like `puts`. This can
break the parsing of output of commands like `brew info --json=v2`.
Let's fix that by capturing the output to stdout, and then printing
those messages to stderr instead (using `opoo` to try to discourage
formula authors from doing this).
- move some things out of `extend` that don't really fit there e.g.
`Module`s that are included but not doing any
overriding/monkeypatching
- move some code into `extend/os` to fix all remaining
`rubocop:todo Homebrew/MoveToExtendOS`s
- remove some unneeded `bundle` skipper code that doesn't really make
sense given our current bottling strategy
- extract some `Pathname` extensions to `extend/pathname` for separate
files
- move a `ENV` `Kernel` extension into `kernel.rb`
- `odeprecate` a seemingly unused backwards compatibility method
- move `readline_nonblock` from a monkeypatch to a
`ReadlineNonblock.read` method as its only used in one place
- fix up a link in documentation
The logic has now been removed in previous commits. This just
removes some references to the `HOMEBREW_INTERNAL_JSON_V3`
environment variable along with reverting the changes to the
`Cachable` class that were originally added in
bd72ec812c3ed656dfcf8e24f77df142a1fe9cc1.
This upgrades `utils/curl.rb` to `typed: strict`, which requires
a number of changes to pass `brew typecheck`. The most
straightforward are adding type signatures to methods, adding type
annotations (e.g., `T.let`) to variables that need them, and ensuring
that methods always use the expected return type.
I had to refactor areas where we call a `Utils::Curl` method and use
array destructuring on a `SystemCommand::Result` return value
(e.g., `output, errors, status = curl_output(...)`), as Sorbet
doesn't understand implicit array conversion. As suggested by Markus,
I've switched these areas to use `#stdout`, `#stderr`, and `#status`.
This requires the use of an intermediate variable (`result`) in some
cases but this was a fairly straightforward substitution.
I also had to refactor how `Cask::URL::BlockDSL::PageWithURL` works.
It currently uses `page.extend PageWithURL` to add a `url` attribute
but this reworks it to subclass `SimpleDelegator` and use an
`initialize` method instead. This achieves the same goal but in a way
that Sorbet can understand.
We've already disabled installing casks/formulae from URLs and we
regularly tell people not to install from paths so let's just deprecate
this behaviour entirely.
Even Homebrew developers do not need to work this way.
- Previously I thought that comments were fine to discourage people from
wasting their time trying to bump things that used `undef` that Sorbet
didn't support. But RuboCop is better at this since it'll complain if
the comments are unnecessary.
- Suggested in https://github.com/Homebrew/brew/pull/18018#issuecomment-2283369501.
- I've gone for a mixture of `rubocop:disable` for the files that can't
be `typed: strict` (use of undef, required before everything else, etc)
and `rubocop:todo` for everything else that should be tried to make
strictly typed. There's no functional difference between the two as
`rubocop:todo` is `rubocop:disable` with a different name.
- And I entirely disabled the cop for the docs/ directory since
`typed: strict` isn't going to gain us anything for some Markdown
linting config files.
- This means that now it's easier to track what needs to be done rather
than relying on checklists of files in our big Sorbet issue:
```shell
$ git grep 'typed: true # rubocop:todo Sorbet/StrictSigil' | wc -l
268
```
- And this is confirmed working for new files:
```shell
$ git status
On branch use-rubocop-for-sorbet-strict-sigils
Untracked files:
(use "git add <file>..." to include in what will be committed)
Library/Homebrew/bad.rb
Library/Homebrew/good.rb
nothing added to commit but untracked files present (use "git add" to track)
$ brew style
Offenses:
bad.rb:1:1: C: Sorbet/StrictSigil: Sorbet sigil should be at least strict got true.
^^^^^^^^^^^^^
1340 files inspected, 1 offense detected
```
This improves the load time of most brew commands. For an example of
one of the simplest commands this speeds up:
Without Bootsnap:
```
$ hyperfine 'git checkout master; brew help' 'git checkout optimise_requires; brew help'
Benchmark 1: git checkout master; brew help
Time (mean ± σ): 525.0 ms ± 35.8 ms [User: 229.9 ms, System: 113.1 ms]
Range (min … max): 465.3 ms … 576.6 ms 10 runs
Benchmark 2: git checkout optimise_requires; brew help
Time (mean ± σ): 383.3 ms ± 25.1 ms [User: 133.0 ms, System: 72.1 ms]
Range (min … max): 353.0 ms … 443.6 ms 10 runs
Summary
git checkout optimise_requires; brew help ran
1.37 ± 0.13 times faster than git checkout master; brew help
```
With Bootsnap:
```
$ hyperfine 'git checkout master; brew help' 'git checkout optimise_requires; brew help'
Benchmark 1: git checkout master; brew help
Time (mean ± σ): 386.0 ms ± 30.9 ms [User: 130.2 ms, System: 93.8 ms]
Range (min … max): 359.5 ms … 469.3 ms 10 runs
Benchmark 2: git checkout optimise_requires; brew help
Time (mean ± σ): 330.2 ms ± 32.4 ms [User: 93.4 ms, System: 73.0 ms]
Range (min … max): 302.9 ms … 413.9 ms 10 runs
Summary
git checkout optimise_requires; brew help ran
1.17 ± 0.15 times faster than git checkout master; brew help
```
This is a follow-up to 484498e. I added loading for tap migration
renames from the API but it apparently only worked for full names.
There was a bug in both the code and the tests which prevented
loading by short names. This fixes those bugs so everything should
be good now.
These would not get loaded from the API at all meaning these
were not loadable by the old names when the core formula or cask
tap was not installed. We assume that most users don't have those
core taps tapped which means this is broken for most everyone.
