Let's start storing `revision` and `pkg_version` for tab runtime
dependencies and use them when available.
When the `revision` is not available, use a conservative approach to
deciding whether dependencies need to be upgrade.
Co-authored-by: Mike McQuaid <mike@mikemcquaid.com>
Ever since we started using this at runtime it's been polluting
the backtrace output. This makes it harder to debug errors and
increases the amount of info users have to paste into the box
when filing an issue.
This is a very direct approach. Essentially, we strip out
everything related to the `sorbet-runtime` gem whenever the top
line in the backtrace is unrelated to sorbet-runtime.
The hope is that this will allow errors related to sorbet to
be diagnosed easily while also reducing the backtrace size
for all other types of errors.
Sometimes it is useful to see the full backtrace though.
For those cases, we include the full backtrace when
`--verbose` is passed in and print a warning that the
Sorbet lines have been removed from the backtrace the
first time they are removed.
Note: This requires gems to be set up so that the call to
`Gem.paths.home` works correctly. For that reason, it must
be included after `utils/gems` which is included in
`standalone/load_path` already.
Download the bottle manifests for the potential formulae we are going to
upgrade and, if they are have all their `runtime_dependencies` versions
currently met, don't try to download the bottle or upgrade the formula.
When we're installing a formula from a bottle, we currently always
upgrade all dependencies in the dependency tree to be safe.
However, if we're installing a bottle and the `runtime_dependencies`
within that bottle's tab all have older or equal versions to those
already installed: we do not need to upgrade these dependencies.
This should help a lot of upgrading a lot of the time, at least for
users using bottles (which is the huge majority).
The only downside or other noticeable change is that this requires us
to download or attempt to download the bottle tab before we compute
the dependencies at installation time.
Co-authored-by: Kevin <apainintheneck@gmail.com>
- warn if running `brew postinstall` explicitly and there's no
`post_install` defined in the formula
- add a `post_install` alias for `brew postinstall` to make life
easier for those jumping between `postinstall` and `post_install` in
e.g. Homebrew development
- refactor `post_install` formula path logic into a new method for
improved readability
- handle the JSON API `post_install` formula path case
The main thing is that this DSL allows us to provide an
interface that can be serialized to the JSON API.
Changes:
- Homebrew::Service
- Adds `#service_name` and `#plist_name` methods
- Each is now included in the `#serialize` method as well
- Eval block on instantiation
- Before we lazy evaluated this but the cost is not significant
and it complicated the code a bunch. This only gets called
during install, when evaluating caveats and in the `brew service`
command. It skips this evaluation if the service block isn't there.
- Add `#command?` helper to avoid `#command.blank?` and `#command.present?`
- Formula
- `#service` now returns a service whenever it's called. This call is
hidden behind a call to `#service?` most of the time anyway so this
should be fine.
- `#plist_name` and `#service_name` now call the methods of the same name
on the service class. This should have already been in the service object
to begin with and keeping these methods here helps preserve backwards
compatibility with people who were overwriting these methods before.
- Caveats
- Prefer `service#command?`
- Add helpers for checking on service commands
- This duplicates some of the work in `brew services`. Maybe we should
merge that repo in at some point.
- Check for installed service at `#plist_name` or `#service_name`. I think
this should be used instead of `Keg#plist_installed?` which checked for any plist file.
We should think about deprecating `#plist_installed?` in the future.
- Stop using `ps aux | grep #{formula.plist_name}` to check for service files
because it was inaccurate (it always returns true on my machine) because the grep
process is started before the ps process.
- Note: The behavior is the same as it was before. This means that caveats
only show up for custom service files on install or if they're already installed.
Otherwise it won't show up in `brew info`. This is because it has to check
first if the service file has been installed.
- Utils::Service
- Add utils for evaluating if a service is installed and running. This duplicates
some of the work already found in `brew services`. We should seriously consider
merging `brew services` with the main brew repo in the future since it's already
tightly coupled to the code in the main repo.
- Formulary.load_formula_from_api
- Be more explicit about which types can be deserialized into run params since
it is now possible for run params to be nil.
- Update and add tests
- roll InfluxDB token (we need to report to a new bucket to fix implicit schema)
- adjust various parameters
- separate default tags and fields
- send more fields and fewer tags (tags should have low cardinality)
- use `--data-binary` to match InfluxDB documentation
- document second precision for greater InfluxDB performance
- pass through tap name, formula/cask name, options separately
- pass `devcmdrun` as a tag
- avoid sending very high-cardinality `OS_VERSION` values
- These are arbitrary length limits that had a load of disables in code.
- The limits were only increasing over time rather than decreasing.
- Fixing the problematic code to be shorter would take a long time for
questionable gain since the problem has been around so long.
- We're not going to make the really long things be any shorter any time soon.
- The instructions in issue 14685 say, pragmatically, "disable all the rubocop
rules we're never going to realistically fix e.g. Metrics/ClassLength". But
that felt like a slippery slope to more _really_ long modules/classes/blocks,
and the limits are here for a reason.
We warn sometimes when we tell people to build from source and it's
not supported but we don't actually warn non-developers when invoking
the various install commands so: let's start doing so.
While we're here, also update the existing messaging to reflect the fact
we're on Mastodon now too and we don't want maintainers being
individually bothered about errors either.
I've tried to balance having messages here vs. having them be so long
at the beginning of installation that they will be missed or be overly
obnoxious.
