Correct and check types on monitoring router and database processes #3572
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
(and of course it fails...)
they were introduced as `queue` module queues, but actually the values (and the annotations used at the other end of the router_starter call) are multiprocessing.Queue (subclasses...) mypy does not typecheck this call... a later PR introduces typeguard here (which detects this) this hasn't caused execution problems - the APIs of the two queue types are close enough - but it is incorrect when trying to understand how pieces of monitoring communicate with each other.
…there are multiple Queue types floating around with very similar interfaces
…otations in modern type syntax
benclifford
added a commit
that referenced
this pull request
Aug 8, 2024
This is in preparation for future type work in the monitoring codebase (for example, see PR #3572). This PR does not claim that the types it is moving around are correct (and PR #3572 contains some instances where the types are incorrect). It is a purely syntactic PR. After this PR, $ git grep '# type:' parsl/monitoring/ returns two remaining comment style annotations, which are 'type: ignore' exclusions not specific types.
|
waiting on #3573 for a bit of syntactic tidyup |
benclifford
added a commit
that referenced
this pull request
Aug 10, 2024
This is in preparation for future type work in the monitoring codebase (for example, see PR #3572). This PR does not claim that the types it is moving around are correct (and PR #3572 contains some instances where the types are incorrect). It is a purely syntactic PR. After this PR, $ git grep '# type:' parsl/monitoring/ returns two remaining comment style annotations, which are 'type: ignore' exclusions not specific types.
khk-globus
approved these changes
Aug 14, 2024
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.
Description
Prior to this PR, the startup code for the monitoring router and database processes had type annotations on queues; but these types were not checked, and were incorrect - they were labelled process-local
Queueinstead of multiprocessing queues. This did not cause much trouble execution- and mypy-wise, as the interfaces of those two classes are similar enough, but it is confusing to read in a part of the codebase that is already confusing (that confusion is probably what lead to the incorrect annotations in the first place...)They were not checked because the informal policy of "internal stuff is checked with mypy, external interfaces are checked with typeguard" works badly here:
The startup methods are launched using multiprocessing.Process, and function invocations are not type-checked by mypy across a multiprocessing Process constructor.
Changed Behaviour
This PR introduces typeguard decorators onto the router and database start methods so that this internal checking happens at runtime. This consequently reveals that the type annotations of these methods are incorrect, and so this PR makes those consequential changes.
Further, generic types (
Queue[MessageType]) are not supported on multiprocessing.Queues before Python 3.12 - so those generic indices are removed from the type annotations. That is unfortunate and weakens in-process static verification - but they could be re-introduced after Parsl drops Python 3.11 support (around 2027 in the present informal support policy)Type of change