Allow cops to invalidate results cache #7496
Merged
Conversation
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
maxh
force-pushed
the
maxh/enable-cache-key
branch
from
ca5e454
to
62bf479
Compare
8 tasks
maxh
force-pushed
the
maxh/enable-cache-key
branch
from
62bf479
to
722c25a
Compare
8 tasks
maxh
force-pushed
the
maxh/enable-cache-key
branch
2 times, most recently
from
5674fea
to
5f930a2
Compare
|
I'm fine with the proposed solution, but I'll defer to @jonas054 to evaluate the implementation. |
jonas054
reviewed
Nov 15, 2019
jonas054
requested changes
Nov 16, 2019
maxh
force-pushed
the
maxh/enable-cache-key
branch
from
5f930a2
to
e7fa9a9
Compare
This was referenced Nov 19, 2019
|
Here's a blog post that goes into more detail about the Rails Engine cops mentioned in the PR description: https://flexport.engineering/isolating-rails-engines-with-rubocop-210feaba3164 When this is merged, I will be able to full upstream those cops. Thanks! |
|
Thanks! |
exterm
reviewed
Nov 25, 2019
| # ResultCache system when those external dependencies change, | ||
| # ie when the ResultCache should be invalidated. | ||
| def external_dependency_checksum | ||
| nil |
There was a problem hiding this comment.
If you run the cop on n files, this method will be called every time, recomputing the checksum n times, correct?
Maybe it would be useful for rubocop to have a "run" context that keeps state across files and is accessible from a cop.
There was a problem hiding this comment.
Good question -- please see #7543. I believe these results should be cached per team/config, so don't need to be recomputed per inspected file.
8 tasks
koic
added a commit
to koic/rubocop-rails
that referenced
this pull request
Apr 9, 2020
…updating schema.rb Fixes rubocop#227. This PR makes `Rails/UniqueValidationWithoutIndex` aware of updating db/schema.rb `Rails/UniqueValidationWithoutIndex` cop needs to know both model and db/schema.rb changes to register an offense. However, with default RuboCop, only changes to the model affect cache behavior. This PR ensures that changes to db/schema.rb affect the cache by overriding the following method: ```ruby # This method should be overridden when a cop's behavior depends # on state that lives outside of these locations: # # (1) the file under inspection # (2) the cop's source code # (3) the config (eg a .rubocop.yml file) # # For example, some cops may want to look at other parts of # the codebase being inspected to find violations. A cop may # use the presence or absence of file `foo.rb` to determine # whether a certain violation exists in `bar.rb`. # # Overriding this method allows the cop to indicate to RuboCop's # ResultCache system when those external dependencies change, # ie when the ResultCache should be invalidated. def external_dependency_checksum nil end ``` https://github.com/rubocop-hq/rubocop/blob/v0.81.0/lib/rubocop/cop/cop.rb#L222-L239 See for more details: rubocop/rubocop#7496
8 tasks
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
Internally at Flexport, we've written a few cops to enforce Rails Engine isolation. They've helped us incrementally modularize our system, and we think they might be valuable to others too. I'm working to open source them now.
For example,
GlobalModelAccessFromEngine(flexport/rubocop-flexport#5) forbids code within Rails Engines from directly accessing global models in the mainapp/directory. Another example isEngineApiBoundary(flexport/rubocop-flexport#6).(Note: we considered upstream these to
rubocop-rails, and @koic suggested we create our gem instead: rubocop/rubocop-rails#152 (comment))These cops read from the filesystem, taking a holistic view of our codebase while they inspect. Unfortunately, this approach does not play nice with the RuboCop results cache. As a workaround, we created a custom
lint.rbthat wrapsrubocopand busts the cache when needed using the--cacheparam. It works ok.But it's not ideal, especially for broader use. This PR aims to fix the issue at its source by allowing cops themselves to invalidate the cache.
Example
Consider the following example:
(1) A filesystem contains these files:
(2) We run
rubocop engines/my_engine/app/services/my_engine/my_service.rband find a violation --service.rbcontainsMyModel.find(123).(2b) During that run, a cached result is stored, keyed by: the inspected source code, the RuboCop config, command-line options, and executable version.
(3) We run the same command again. Cache is hit, same violation shown. All good.
(4) We move the model file into the engine. So now
app/models/my_model.rbno longer exists and we haveengines/my_engine/app/models/my_engine/my_model.rb.(5) We run the same RuboCop command again. Because none of the cache key inputs changed, the same violation is shown again. This is incorrect.
(6) We run the same command again with
--cache falseand see that the violation no longer exists. This is correct.Implementation overview
This PR allows cops to define an
external_dependency_checksummethod that busts the cache when their external dependencies change. Cops are responsible for computing their own checksum however they deem appropriate.Among existing cops, I believe there are no use cases for this method. For the cops we've written, there are two types of external dependencies: (1) the presence or absence of certain files and/or (2) the contents of certain files.
Discussion
I recognize that this is perhaps an unusual use case for RuboCop. As an alternative, we could create our own repo with these cache-unfriendly cops and package them with a custom runner script that does cache busting inside the script. But that seems suboptimal. And I suspect other cops may make use of this feature in the future as well.
Feedback welcome! Thanks!
Before submitting the PR make sure the following are checked:
[Fix #issue-number](if the related issue exists).master(if not - rebase it).bundle exec rake default. It executes all tests and RuboCop for itself, and generates the documentation.