- This cop checks for the use of `FileUtils.rm_rf` and suggests using
`FileUtils.rm_r` because we should know if we couldn't delete a
thing for some reason, not just force it.
- This can be either a String or a Pathname, per
the part of the `noseyparker` test that failed
(in a different part of the test, the command is
passed as a string).
```
==> Testing noseyparker
==> /opt/homebrew/Cellar/noseyparker/0.18.1/bin/noseyparker -V
Error: noseyparker: failed
An exception occurred within a child process:
TypeError: Parameter 'cmd': Expected type String, got type Pathname with value #<Pathname:/opt/homebrew/Ce...ps://github.com/Homebrew/brew>
Caller: /opt/homebrew/Library/Taps/homebrew/homebrew-core/Formula/n/noseyparker.rb:35
```
- I was on a plane and this failed with:
```
expected block to not output to stderr, but output
"curl: (6) Could not resolve host: spdx.github.io\nWarning: Failed to fetch SBOM schema, cannot perform SBOM validation!
Warning: Could not fetch JSON schema to validate SBOM!\nWarning: SBOM is not valid, not writing to disk!"
```
- I put a debugger call in the test that was failing.
- Running the install command at that debug prompt and lo,
the typing bug was staring me in the face:
```
Error: An exception occurred within a child process:
TypeError: Parameter 'ref_type': Expected type T.nilable(String), got type Symbol with value :branch
Caller: /opt/homebrew/Library/Homebrew/vendor/bundle/ruby/3.3.0/gems/sorbet-runtime-0.5.11471/lib/types/private/methods/call_validation.rb:215
Definition: /opt/homebrew/Library/Homebrew/unpack_strategy.rb:102 (UnpackStrategy.detect)
```
- This should fix https://github.com/Homebrew/homebrew-core/actions/runs/9839505520, that is:
```
Parameter 'tag': Expected type Float, got type String with value "2024-07-05"
Caller: /home/linuxbrew/.linuxbrew/Homebrew/Library/Homebrew/vendor/bundle/ruby/3.3.0/gems/sorbet-runtime-0.5.11473/lib/types/private/methods/call_validation.rb:215
Definition: /home/linuxbrew/.linuxbrew/Homebrew/Library/Homebrew/dev-cmd/bump-formula-pr.rb:445 (Homebrew::DevCmd::BumpFormulaPr#fetch_resource_and_forced_version)
```
- I think I read https://sorbet.org/docs/sigs#rest-parameters too literally,
yes Sorbet expands the kwargs into a hash type, but it's the value type that
needs specifying in the sig (which for us is `String`).
- Some download locations return a non-standard formatting of date string for the `Last-Modified` header.
This causes `Time.parse` to blow up. The user sees `error: argument out of range`.
- In this commit we handle the error and return nil, which `filter_map` (equivalent to `.map.compact`) gets rid of and then `time.last` is as normal.
- Fixes https://github.com/Homebrew/brew/issues/ 17556.
- Not only `false` to `true` but `true` to `strict`.
- Only humans every run this, but our goal is to increase the
typechecking in our files to get to `strict` everywhere so
let's make that easy to remember to do.
- Both `from` and `to` are now separate keyword arguments
in a bunch of places, not part of `args`.
- When we switched this around, we didn't realize this
method needed updating to correctly construct the time
range query.
- This led to further inaccurate counts in `brew contributions`
for reviews, since `from` and `to` are not valid search qualifiers
for the GitHub PR search APIs.
- This was broken (I did have a commit SHA for the breakage but I can't find it now) since `from` and `args.from` are different variables (one can be nil, the other has a default value).
- So it was reporting very high counts because, despite the message, the `from` restriction was not being passed to `count_repo_commits`.
- Only two audits were using this: `audit_keg_only_reason` and `audit_text`,
and they weren't using any of its text processing methods, so there's little
reason to keep it around.
- The "`keg_only_reason` shouldn't contain 'HOMEBREW_PREFIX'" audit can easily
be replaced with a RuboCop since that's "just" text parsing.
- The "tests should invoke binaries with `bin/<command>`" audit had to stay as
a FormulaAudit because it requires accessing attributes about the Formula
like its name, aliases, which RuboCop can't get to, but it was easy to move the
singular "read the text in the file" line from `FormulaTextAuditor`.
- It was suggested in https://github.com/Homebrew/brew/issues/ 16733 that
`brew desc <formula_or_cask>` should work like `brew info <formula_or_cask>`
and print the description of the package without needing `--eval-all`.
- Looking at the code, it seems like it's only searching that needs
`--eval-all`, so limit the check to that.
Before:
```shell
$ brew desc hello
Error: `brew desc` needs `--eval-all` passed or `HOMEBREW_EVAL_ALL` set!
```
After:
```shell
$ brew desc hello
hello: Program providing model for GNU coding standards and practices
$ brew desc --search hello
Error: Invalid usage: `brew desc --search` needs `--eval-all` passed or `HOMEBREW_EVAL_ALL` set!
$ brew desc --search --eval-all hello
==> Formulae
dsh: Dancer's shell, or distributed shell
hello: Program providing model for GNU coding standards and practices
```
- This doesn't work still, but now for new reasons.
```
NoMethodError:
undefined method `b' for #<Binding:0x0000000107a7e088>
binding.b
^^
# ./test/migrator_spec.rb:72:in `block (3 levels) in <top (required)>'
```
and if I add `require "debug"` to the spec_helper again it gives:
```
LoadError:
cannot load such file -- debug
```
but, doing a `require "debug"; binding.b` in `migrator_spec` _does_ work.
Where is the require coming from where it works some of the time but not through all of the layers we have?
- This is cleaner than vendoring a whole bunch of new gems and pinning `psych`.
- Thanks for the pointer, Bo!
- It doesn't work, though?
```
❯ brew tests --only=migrator --debug
Error: cannot load such file -- debug/debug.so
Warning: Removed Sorbet lines from backtrace!
Rerun with `--verbose` to see the original backtrace
/opt/homebrew/Library/Homebrew/vendor/portable-ruby/3.1.4/lib/ruby/gems/3.1.0/gems/debug-1.6.3/lib/debug/frame_info.rb:16:in `require'
/opt/homebrew/Library/Homebrew/vendor/portable-ruby/3.1.4/lib/ruby/gems/3.1.0/gems/debug-1.6.3/lib/debug/frame_info.rb:16:in `rescue in <module:DEBUGGER__>'
/opt/homebrew/Library/Homebrew/vendor/portable-ruby/3.1.4/lib/ruby/gems/3.1.0/gems/debug-1.6.3/lib/debug/frame_info.rb:13:in `<module:DEBUGGER__>'
/opt/homebrew/Library/Homebrew/vendor/portable-ruby/3.1.4/lib/ruby/gems/3.1.0/gems/debug-1.6.3/lib/debug/frame_info.rb:3:in `<top (required)>'
/opt/homebrew/Library/Homebrew/vendor/portable-ruby/3.1.4/lib/ruby/gems/3.1.0/gems/debug-1.6.3/lib/debug/session.rb:31:in `require_relative'
/opt/homebrew/Library/Homebrew/vendor/portable-ruby/3.1.4/lib/ruby/gems/3.1.0/gems/debug-1.6.3/lib/debug/session.rb:31:in `<top (required)>'
/opt/homebrew/Library/Homebrew/vendor/portable-ruby/3.1.4/lib/ruby/gems/3.1.0/gems/debug-1.6.3/lib/debug.rb:3:in `require_relative'
/opt/homebrew/Library/Homebrew/vendor/portable-ruby/3.1.4/lib/ruby/gems/3.1.0/gems/debug-1.6.3/lib/debug.rb:3:in `<top (required)>'
/opt/homebrew/Library/Homebrew/dev-cmd/tests.rb:48:in `require'
/opt/homebrew/Library/Homebrew/dev-cmd/tests.rb:48:in `run'
/opt/homebrew/Library/Homebrew/brew.rb:89:in `<main>'
```
- Byebug was introduced in [2020](https://github.com/Homebrew/brew/pull/7577) for hooking into tests for debugging.
- It does not work anymore in so far as it does not stop at breakpoints when following the instructions to trigger them in tests.
- Spoiler: they aren't.
- As part of Volunteer Month at work I introduced `hharen` to contributing to
Homebrew, `cd $(brew --repo)`, `brew tests`, `brew style`, etc.
- We started to write a RuboCop for this. But my current thinking is that this
test might be sufficient since it might be easier to notice "oop, a test is
failing and I've added a new envvar" and re-alphabetize it than to write a
RuboCop linter for it to do the one-time autofix. Considering how little this
gets changed?
Co-authored-by: Hana <hharen@github.com>
- The preferred way of doing RBI generation is via Tapioca. So I am
trying to stop being intimidated by it, by learning how it works.
- This is very WIP still, currently failing with the following message
because the `module` name is missing in the generated RBI file.
```
There are parse errors in the generated RBI files.
Errors:
sorbet/rbi/dsl/tty.rbi:8: unexpected token tNL (2001)
sorbet/rbi/dsl/tty.rbi:64: unexpected token "end" (2001)
```
- Some repositories occasionally change their licenses. For example they
release a version of the software with one license and then decide to change
the license later.
- Now that `?ref=` is a parameter to the GitHub Repositories License API,
we can use that in the license audit to check if the license of the specific
release matches the one declared in the formula.
- We're moving from `depends_on "python-lxml"` to `resource "lxml" ...` as part
of the new Python vendoring plan.
- For `resource "lxml"` to work, `uses_from_macos "libxml2"` and
`uses_from_macos "libxslt"` are needed in the formulae.
- This new RuboCop rule enforces that a formula including the "lxml" resource
also include those dependencies.
- This diagram was really old and didn't reflect the current practices
(a lot more things are automated now, hurray).
- Anything we replace it with will get outdated fast too.
- Also delete the diagramming docs since I don't believe we have made
another diagram since this one which got out of date and forgotten
about since it was a DrawIO SVG and not easily editable as code.
- If we want diagrams these days, we can use Mermaid which is
diagrams-as-code.
- We don't need this blanket enabled, we only have a few docs where it actually matters that there's a description, it's just cruft having to invent descriptions.
- Yes, we have yet another `.rubocop.yml` but in-line exclusions in
`<span class="display:none;">` in the Markdown seemed more ugly.
- This needed tweaks to `brew style` to make it read the new config
file when we're scanning docs.
