Add Ty type checking [awaiting Ty beta] #458
Conversation
Still some failures I need to track down, but mostly working
…and fix ty issues
WalkthroughThis update removes the use of the Pyright type checker in favor of the "ty" type checker throughout the codebase, including CI scripts, documentation, and dependency files. It also replaces a custom function for exhaustive enum checks with the standard Changes
Sequence Diagram(s)sequenceDiagram
participant Dev as Developer
participant CI as CI Pipeline
participant Ty as Ty Type Checker
Dev->>CI: Push code / open PR
CI->>Ty: Run "uvx ty check"
Ty-->>CI: Report type checking results
CI-->>Dev: Pass/fail status
Estimated code review effort🎯 3 (Moderate) | ⏱️ ~20 minutes Poem
Note ⚡️ Unit Test Generation is now available in beta!Learn more here, or try it out under "Finishing Touches" below. 📜 Recent review detailsConfiguration used: CodeRabbit UI 📒 Files selected for processing (1)
🚧 Files skipped from review as they are similar to previous changes (1)
⏰ Context from checks skipped due to timeout of 90000ms. You can increase the timeout in your CodeRabbit configuration to a maximum of 15 minutes (900000ms). (5)
Thanks for using CodeRabbit! It's free for OSS, and your support helps us grow. If you like it, consider giving us a shout-out. 🪧 TipsChatThere are 3 ways to chat with CodeRabbit:
SupportNeed help? Create a ticket on our support page for assistance with any issues or questions. Note: Be mindful of the bot's finite context window. It's strongly recommended to break down tasks such as reading entire modules into smaller chunks. For a focused discussion, use review comments to chat about specific files and their changes, instead of using the PR comments. CodeRabbit Commands (Invoked using PR comments)
Other keywords and placeholders
CodeRabbit Configuration File (
|
There was a problem hiding this comment.
Actionable comments posted: 4
🧹 Nitpick comments (6)
libs/core/kiln_ai/datamodel/basemodel.py (1)
419-421: Drop the inline annotation instead of silencing the checker
tyflags the free variablechild_classbecause it’s resolved only at runtime.
Rather than adding# type: ignore, omit the return annotation in the function header and rely on the explicit__annotations__assignment just below, which already provides the correct runtime/type-introspection information.- def child_method(self, readonly: bool = False) -> list[child_class]: # type: ignore + # Return annotation is injected after definition to avoid runtime-only symbol issues + def child_method(self, readonly: bool = False): return child_class.all_children_of_parent_path(self.path, readonly=readonly)This keeps the signature clean, avoids duplicate annotation writes, and removes the need for a blanket ignore.
app/desktop/studio_server/finetune_api.py (2)
223-230: Consider removing redundant provider validation.The manual provider validation check (lines 223-227) is now redundant since
base_provider_from_str_idperforms the same validation. However, the error messages differ:
- Manual check: "Fine tune provider '{provider}' not found"
- Helper function: "Invalid provider '{provider}'. Valid providers are: {valid_providers}"
If the more informative error message from the helper function is preferred, the manual check can be removed.
- if finetune.provider not in finetune_registry: - raise HTTPException( - status_code=400, - detail=f"Fine tune provider '{finetune.provider}' not found", - ) finetune_adapter = base_provider_from_str_id(finetune.provider)The usage of the helper function itself is correct and improves code consistency.
369-374: Consider removing redundant provider validation.Similar to the earlier endpoint, there's redundant validation logic here. The manual check (lines 369-373) and the helper function perform the same validation with different error messages:
- Manual check: "Fine tune provider '{provider}' not found"
- Helper function: "Invalid provider '{provider}'. Valid providers are: {valid_providers}"
The helper function provides better UX by listing valid options.
- if request.provider not in finetune_registry: - raise HTTPException( - status_code=400, - detail=f"Fine tune provider '{request.provider}' not found", - ) finetune_adapter_class = base_provider_from_str_id(request.provider)The helper function usage is correct and improves consistency.
README.md (1)
20-21: Table header is now blank – give the first column a label for clarity & accessibilityThe first header cell was emptied (
| |). Screen-readers and MD renderers will now announce this column as “blank”, which hurts accessibility and makes the source harder to skim. A succinct label such asSectionorCategorykeeps the intent clear.-| | | +| Section | |libs/core/pyproject.toml (1)
44-45: Duplicate dev-dependency declarations – consider hoisting to workspace root
tyis now declared in bothlibs/coreandlibs/server. If another sub-package is added we’ll have a third copy. Hoist common dev deps to the workspace-rootpyproject.tomlunder a shareddevgroup to avoid version drift.CONTRIBUTING.md (1)
69-70: Give the exact VS Code extension identifier for TyBeginners will search unsuccessfully for “Ty”. Add the marketplace ID so the extension can be installed with one click, e.g.:
- Ty (astral type checker) + Ty (astral type checker) – `astral-sh.ty-vscode`
📜 Review details
Configuration used: CodeRabbit UI
Review profile: CHILL
Plan: Pro
⛔ Files ignored due to path filters (3)
libs/core/uv.lockis excluded by!**/*.locklibs/server/uv.lockis excluded by!**/*.lockuv.lockis excluded by!**/*.lock
📒 Files selected for processing (29)
.github/workflows/build_and_test.yml(1 hunks)AGENTS.md(1 hunks)CONTRIBUTING.md(1 hunks)README.md(1 hunks)app/desktop/studio_server/finetune_api.py(5 hunks)app/desktop/studio_server/provider_api.py(5 hunks)app/desktop/studio_server/test_finetune_api.py(1 hunks)checks.sh(1 hunks)libs/core/kiln_ai/adapters/adapter_registry.py(2 hunks)libs/core/kiln_ai/adapters/chat/chat_formatter.py(3 hunks)libs/core/kiln_ai/adapters/eval/base_eval.py(2 hunks)libs/core/kiln_ai/adapters/eval/registry.py(2 hunks)libs/core/kiln_ai/adapters/fine_tune/dataset_formatter.py(2 hunks)libs/core/kiln_ai/adapters/model_adapters/base_adapter.py(1 hunks)libs/core/kiln_ai/adapters/model_adapters/litellm_adapter.py(5 hunks)libs/core/kiln_ai/adapters/model_adapters/test_litellm_adapter.py(1 hunks)libs/core/kiln_ai/adapters/parsers/parser_registry.py(2 hunks)libs/core/kiln_ai/adapters/parsers/request_formatters.py(2 hunks)libs/core/kiln_ai/adapters/parsers/test_parser_registry.py(1 hunks)libs/core/kiln_ai/adapters/parsers/test_request_formatters.py(1 hunks)libs/core/kiln_ai/adapters/prompt_builders.py(2 hunks)libs/core/kiln_ai/adapters/provider_tools.py(1 hunks)libs/core/kiln_ai/datamodel/basemodel.py(1 hunks)libs/core/kiln_ai/datamodel/eval.py(3 hunks)libs/core/kiln_ai/datamodel/task_output.py(2 hunks)libs/core/kiln_ai/utils/exhaustive_error.py(0 hunks)libs/core/pyproject.toml(1 hunks)libs/server/pyproject.toml(1 hunks)pyproject.toml(2 hunks)
💤 Files with no reviewable changes (1)
- libs/core/kiln_ai/utils/exhaustive_error.py
🧰 Additional context used
📓 Path-based instructions (2)
**/*.py
📄 CodeRabbit Inference Engine (.cursor/rules/project.mdc)
**/*.py: Always assume pydantic 2 (not pydantic 1)
The project supports Python 3.10 and above
Files:
libs/core/kiln_ai/adapters/parsers/test_request_formatters.pylibs/core/kiln_ai/datamodel/task_output.pylibs/core/kiln_ai/adapters/parsers/request_formatters.pylibs/core/kiln_ai/adapters/adapter_registry.pylibs/core/kiln_ai/adapters/eval/registry.pylibs/core/kiln_ai/adapters/eval/base_eval.pyapp/desktop/studio_server/test_finetune_api.pylibs/core/kiln_ai/adapters/parsers/parser_registry.pylibs/core/kiln_ai/adapters/parsers/test_parser_registry.pylibs/core/kiln_ai/adapters/provider_tools.pylibs/core/kiln_ai/adapters/model_adapters/test_litellm_adapter.pylibs/core/kiln_ai/datamodel/basemodel.pylibs/core/kiln_ai/adapters/prompt_builders.pyapp/desktop/studio_server/provider_api.pylibs/core/kiln_ai/datamodel/eval.pylibs/core/kiln_ai/adapters/model_adapters/litellm_adapter.pylibs/core/kiln_ai/adapters/fine_tune/dataset_formatter.pylibs/core/kiln_ai/adapters/model_adapters/base_adapter.pylibs/core/kiln_ai/adapters/chat/chat_formatter.pyapp/desktop/studio_server/finetune_api.py
**/test_*.py
📄 CodeRabbit Inference Engine (.cursor/rules/project.mdc)
**/test_*.py: Always use pytest for tests in Python code
Test brevity is important. Use approaches for re-use and brevity including using fixtures for repeated code, and using pytest parameterize for similar tests
Files:
libs/core/kiln_ai/adapters/parsers/test_request_formatters.pyapp/desktop/studio_server/test_finetune_api.pylibs/core/kiln_ai/adapters/parsers/test_parser_registry.pylibs/core/kiln_ai/adapters/model_adapters/test_litellm_adapter.py
🧠 Learnings (7)
AGENTS.md (1)
Learnt from: CR
PR: Kiln-AI/Kiln#0
File: .cursor/rules/project.mdc:0-0
Timestamp: 2025-07-21T15:28:29.450Z
Learning: To run backend / Python tests, run uv run python3 -m pytest . -s -v or a variant of it to run individual tests
.github/workflows/build_and_test.yml (1)
Learnt from: CR
PR: Kiln-AI/Kiln#0
File: .cursor/rules/project.mdc:0-0
Timestamp: 2025-07-21T15:28:29.450Z
Learning: To run backend / Python tests, run uv run python3 -m pytest . -s -v or a variant of it to run individual tests
libs/core/kiln_ai/datamodel/task_output.py (3)
Learnt from: leonardmq
PR: #418
File: libs/core/kiln_ai/adapters/ml_model_list.py:0-0
Timestamp: 2025-07-16T09:37:39.816Z
Learning: The glm_z1_rumination_32b_0414 model was intentionally removed from the built_in_models list due to output formatting issues: output was duplicated in both output and reasoning fields, and contained random internal JSON in the output. This model should not be re-added without addressing these formatting problems.
Learnt from: leonardmq
PR: #418
File: libs/core/kiln_ai/adapters/ml_model_list.py:0-0
Timestamp: 2025-07-16T09:37:39.816Z
Learning: The glm_z1_rumination_32b_0414 model was intentionally removed from the built_in_models list due to output formatting issues: output was duplicated in both output and reasoning fields, and contained random internal JSON in the output. This model should not be re-added without addressing these formatting problems.
Learnt from: leonardmq
PR: #313
File: libs/core/kiln_ai/adapters/extraction/gemini_extractor.py:86-88
Timestamp: 2025-05-15T17:16:16.742Z
Learning: In Kiln's extraction configuration, Pydantic automatically coerces string keys (e.g., "document") into Kind enum values when validating dictionary fields typed as dict[Kind, str], so no manual conversion is needed when accessing these values.
CONTRIBUTING.md (1)
Learnt from: CR
PR: Kiln-AI/Kiln#0
File: .cursor/rules/project.mdc:0-0
Timestamp: 2025-07-21T15:28:29.450Z
Learning: Applies to **/*.{js,jsx} : We always use TypeScript, not JavaScript
checks.sh (1)
Learnt from: CR
PR: Kiln-AI/Kiln#0
File: .cursor/rules/project.mdc:0-0
Timestamp: 2025-07-21T15:28:29.450Z
Learning: To run backend / Python tests, run uv run python3 -m pytest . -s -v or a variant of it to run individual tests
libs/core/kiln_ai/adapters/provider_tools.py (1)
Learnt from: leonardmq
PR: #315
File: libs/core/kiln_ai/adapters/extractors/registry.py:3-3
Timestamp: 2025-06-01T17:11:57.451Z
Learning: The from google import genai import is valid and working correctly in the Kiln AI codebase, despite static analysis tools potentially flagging it as an error.
libs/core/kiln_ai/datamodel/eval.py (1)
Learnt from: leonardmq
PR: #313
File: libs/core/kiln_ai/adapters/extraction/gemini_extractor.py:86-88
Timestamp: 2025-05-15T17:16:16.742Z
Learning: In Kiln's extraction configuration, Pydantic automatically coerces string keys (e.g., "document") into Kind enum values when validating dictionary fields typed as dict[Kind, str], so no manual conversion is needed when accessing these values.
🧬 Code Graph Analysis (2)
libs/core/kiln_ai/adapters/parsers/test_parser_registry.py (1)
libs/core/kiln_ai/adapters/parsers/parser_registry.py (1)
model_parser_from_id(8-20)
app/desktop/studio_server/finetune_api.py (2)
libs/core/kiln_ai/adapters/fine_tune/base_finetune.py (3)
BaseFinetuneAdapter(40-182)FineTuneParameter(21-29)FineTuneStatus(12-18)app/web_ui/src/lib/types.ts (1)
FineTuneParameter(19-19)
⏰ Context from checks skipped due to timeout of 90000ms. You can increase the timeout in your CodeRabbit configuration to a maximum of 15 minutes (900000ms). (5)
- GitHub Check: Build Desktop Apps (ubuntu-22.04-arm)
- GitHub Check: Build Desktop Apps (macos-latest)
- GitHub Check: Build Desktop Apps (ubuntu-22.04)
- GitHub Check: Build Desktop Apps (windows-latest)
- GitHub Check: Build Desktop Apps (macos-13)
🔇 Additional comments (38)
app/desktop/studio_server/test_finetune_api.py (1)
369-370: LGTM: Test correctly updated to match new error message format.The change from exact string match to substring containment check is appropriate and aligns with the centralized provider validation implementation mentioned in the AI summary. The test still validates the core behavior (400 status code for invalid provider) while accommodating the new error message format that includes a list of valid providers.
app/desktop/studio_server/provider_api.py (4)
28-28: LGTM: Proper import of standardassert_neverfunction.Good practice to use the standard library's
assert_neverfromtyping_extensionsinstead of a custom function for exhaustive enum checking.
48-48: LGTM: Correct exception import.The change from
requests.exceptions.ConnectionErrortorequests.ConnectionErroris the proper way to import and catch this exception type.
311-311: LGTM: Proper use ofassert_neverfor exhaustive enum checking.The replacement of the custom exhaustive enum error function with the standard
assert_neverfollows Python best practices and provides better type safety.
366-367: LGTM: Consistent use ofassert_neverwith improved comment.Good replacement of the custom function with standard
assert_never, and the comment clearly explains the purpose of the exhaustive check.app/desktop/studio_server/finetune_api.py (2)
8-12: LGTM! Import addition supports proper type hinting.The addition of
BaseFinetuneAdapterimport is necessary for the new helper function's return type annotation.
296-297: LGTM! Improved error handling for invalid provider IDs.This replacement adds proper validation and error handling where there previously would have been a KeyError for invalid provider IDs. The helper function will now return a clear 400 error instead of a generic 500 error.
README.md (1)
25-25: Verify the new Ty badge URLThe badge now points to a personal-gist URL:
https://gist.githubusercontent.com/scosman/.../ty_badge.jsonGists are easy to delete or move, which would break the badge.
Consider hosting the JSON in the repo (e.g. under.badges/ty.json) or on a project-controlled URL to ensure long-term stability.AGENTS.md (1)
15-15: Confirm CI treats Ty warnings as failures if desired
uvx ty checkexits with0when it emits warnings but no errors.
Pyright previously allowed--level errorto make the pipeline strict.
If you relied on that behaviour, add Ty’s flag:uvx ty check --strict(or update the docs to mention that warnings are acceptable).
libs/core/kiln_ai/adapters/parsers/test_parser_registry.py (1)
16-19: LGTM! Test correctly updated for assert_never behavior.The test properly expects
AssertionErrorwith the "Expected code to be unreachable" message, which aligns with theassert_neverimplementation in the correspondingmodel_parser_from_idfunction..github/workflows/build_and_test.yml (1)
40-40: LGTM! Correctly updated to use Ty type checker.The workflow properly switches from
uv run pyright .touvx ty check, aligning with the PR's objective to migrate from Pyright to the Ty type checker.libs/core/kiln_ai/adapters/parsers/test_request_formatters.py (1)
75-75: LGTM! Test correctly updated for assert_never behavior.The test properly expects
AssertionErrorwith regex match for "Expected code to be unreachable", which is consistent with theassert_neverimplementation pattern used throughout this refactor.libs/core/kiln_ai/datamodel/task_output.py (2)
6-6: LGTM! Correctly imports assert_never for exhaustive enum handling.The import properly adds
assert_neverfromtyping_extensions, replacing the custom exhaustive error function with the standard typing utility.
44-44: LGTM! Properly replaced custom error with assert_never.The default case correctly uses
assert_never(rating_type)instead of the customraise_exhaustive_enum_error, maintaining exhaustive enum checking with the standard typing utility.libs/core/kiln_ai/adapters/eval/registry.py (2)
1-1: LGTM! Correctly imports assert_never for exhaustive enum handling.The import properly adds
assert_neverfromtyping_extensionsto replace the custom exhaustive error function.
17-17: LGTM! Properly replaced custom error with assert_never.The default case correctly uses
assert_never(eval_config_type)instead of the custom error function, maintaining exhaustive enum checking with the standard typing utility.libs/core/kiln_ai/adapters/parsers/parser_registry.py (1)
1-1: Excellent migration to standard library functionGreat change replacing the custom
raise_exhaustive_enum_errorwithassert_neverfromtyping_extensions. This provides better type checking support and is more idiomatic Python code.Also applies to: 20-20
libs/core/kiln_ai/adapters/adapter_registry.py (1)
3-4: LGTM - Consistent with codebase refactorGood migration to
assert_neverfor exhaustive enum handling, consistent with the broader codebase changes.Also applies to: 200-200
checks.sh (1)
16-16: Improved script clarity and ty integrationGood improvements to the script:
- More descriptive header separating linting/formatting from type checking
- Proper integration of the ty type checker with
uvx ty check- Clear separation of concerns in the output
Also applies to: 21-23
libs/core/kiln_ai/adapters/provider_tools.py (1)
387-388: Good migration with helpful commentExcellent change to
assert_neverwith an updated comment that clearly explains the type checking benefit. This maintains exhaustive enum handling while using standard library functionality.libs/core/kiln_ai/adapters/prompt_builders.py (1)
3-4: Consistent migration completing the refactorPerfect completion of the
assert_nevermigration pattern. The comment correctly explains the type checking benefit, and this change maintains consistency across the entire codebase.Also applies to: 429-429
libs/core/kiln_ai/adapters/fine_tune/dataset_formatter.py (1)
8-8: LGTM: Correct adoption of standard exhaustive enum checking.The replacement of the custom
raise_exhaustive_enum_errorfunction withassert_neverfromtyping_extensionsis the correct approach for exhaustive enum handling in Python. This change provides better type checking support and follows standard typing practices.Also applies to: 97-97
libs/core/kiln_ai/adapters/parsers/request_formatters.py (1)
4-4: LGTM: Consistent adoption of standard exhaustive enum checking.The change from
raise_exhaustive_enum_errortoassert_neveris correctly implemented and consistent with the broader refactor. This improves type checking support while maintaining the same runtime behavior.Also applies to: 41-41
libs/core/kiln_ai/adapters/eval/base_eval.py (1)
5-5: LGTM: Proper implementation of standard exhaustive enum checking.The replacement of
raise_exhaustive_enum_errorwithassert_neveris correctly implemented. This change standardizes exhaustive enum handling using the typing utilities while maintaining the same logical behavior.Also applies to: 162-162
libs/core/kiln_ai/adapters/model_adapters/test_litellm_adapter.py (1)
346-347: LGTM: Test correctly updated to match assert_never behavior.The test has been properly updated to expect
AssertionErrorwith the message "Expected code to be unreachable", which is the correct behavior whenassert_neveris called. This change aligns with the implementation switch from custom error handling to standard typing utilities.pyproject.toml (1)
46-47: LGTM: Appropriate ty configuration with sensible exclusions.The
tyconfiguration correctly excludes test files, build artifacts, web UI components, and virtual environments from type checking. This is a sensible setup that focuses type checking on the core source code.libs/core/kiln_ai/adapters/model_adapters/base_adapter.py (1)
259-266: LGTM: Improved serialization consistencyThe updated logic now consistently preserves string inputs/outputs as-is while serializing all non-string types to JSON. This is more robust than the previous approach that only handled dictionaries, and aligns with similar changes across the codebase for consistent serialization behavior.
libs/core/kiln_ai/adapters/model_adapters/litellm_adapter.py (5)
2-2: LGTM: Correct import updates for type safety improvementsThe addition of
castandassert_neverimports aligns with the broader refactor to standardize exhaustive enum handling and resolve type inference issues.Also applies to: 7-7
155-162: LGTM: Type cast resolves inference issueThe cast is appropriately documented as a workaround for a specific Ty type checker issue. This ensures type safety while the underlying issue is resolved.
195-195: LGTM: Standardized exhaustive enum handlingReplacing the custom error function with
assert_neverfollows the standard typing pattern for exhaustive enum matching and improves type safety.
308-310: LGTM: Type cast improves match statement inferenceThe cast to
ModelProviderNamebefore the match statement helps the type checker understand the exhaustive nature of the enum matching.
344-344: LGTM: Consistent exhaustive enum handlingThis completes the standardization of exhaustive enum error handling throughout the file, replacing custom error functions with the standard
assert_neverpattern.libs/core/kiln_ai/datamodel/eval.py (3)
3-3: LGTM: Standard import updates for type safetyThe addition of
castandassert_neverimports supports the type safety improvements and exhaustive enum handling standardization.Also applies to: 6-6
138-147: LGTM: Improved type safety and defensive programmingThe type cast resolves inference issues with the new type checker, and the added guard clause for
Noneparent_eval_config improves runtime safety by preventing potential errors downstream.
196-197: LGTM: Well-documented exhaustive enum handlingThe replacement with
assert_neverfollows the standard pattern, and the comment clearly explains that this case should never occur due to the exhaustive enum matching above.libs/core/kiln_ai/adapters/chat/chat_formatter.py (3)
8-9: LGTM: Standard import for exhaustive enum handlingThe addition of
assert_neverimport supports the standardization of exhaustive enum handling across the codebase.
219-219: LGTM: Standardized exhaustive enum handlingReplacing the custom error function with
assert_neverfollows the standard typing pattern for exhaustive enum matching and improves type safety.
231-234: LGTM: Improved serialization logic clarityThe inverted logic is more readable and follows a clear pattern: preserve strings as-is, serialize everything else to JSON. This aligns with similar changes in other modules for consistent serialization behavior.
| "ty>=0.0.1a11", | ||
| ] |
There was a problem hiding this comment.
💡 Verification agent
🧩 Analysis chain
Alpha pre-release pin of Ty may cause unexpected breakage
ty>=0.0.1a11 is still an a (alpha) build. Upstream has been shipping breaking API changes every few days. For CI stability you may want to:
- Pin to the exact version used in CI (e.g.
==0.0.1a11) and bump deliberately, or - Track the latest pre-release but add a dependabot rule so breakage is caught quickly.
- "ty>=0.0.1a11",
+ # Pin exact version for reproducible dev installs
+ "ty==0.0.1a11",🏁 Script executed:
#!/usr/bin/env bash
set -euo pipefail
echo "Fetching release tags for 'ty' from PyPI..."
curl -s https://pypi.org/pypi/ty/json \
| jq -r '.releases | keys[]' \
| sort -V \
| tail -n 10Length of output: 301
Pin Ty pre-release dependency to avoid unexpected breakage
PyPI now hosts Ty up through 0.0.1a16, so using
"ty>=0.0.1a11"will automatically pull in 0.0.1a12–a16 (or future alphas) and risks CI instability when upstream makes breaking API changes. For reproducible installs, please choose one of:
- Pin the exact version used in CI:
- "ty>=0.0.1a11", + # Pin exact alpha version for reproducibility + "ty==0.0.1a11",
- Alternatively, lock to the current alpha series to prevent auto-upgrades beyond a11:
- "ty>=0.0.1a11", + # Allow only 0.0.1a11 installs (no newer alphas) + "ty>=0.0.0.1a11,<0.0.1a12",
- Or continue tracking the latest pre-release but add a Dependabot or Renovate rule to catch breaking changes immediately.
🤖 Prompt for AI Agents
In libs/server/pyproject.toml around lines 37 to 38, the dependency on Ty is
specified as "ty>=0.0.1a11", which allows automatic upgrades to newer
pre-release versions that may introduce breaking changes. To fix this, pin the
Ty dependency to a specific version used in CI (e.g., "ty==0.0.1a11") or
restrict it to the current alpha series (e.g., "ty>=0.0.1a11,<0.0.2") to ensure
reproducible installs and prevent unexpected breakage.
|
Should we also disable the Python language server in the They recommend disabling language server in their Cursor / VS Code extension page: We might also want to add a |
|
@CodeRabbit review |
✅ Actions performedReview triggered.
|
There was a problem hiding this comment.
Actionable comments posted: 0
🧹 Nitpick comments (2)
.github/workflows/format_and_lint.yml (2)
37-37: Consider caching theuv syncinstallation step
uv syncre-resolves and installs all dependencies on every run. Adding a cache keyed on the lock-file hash (e.g..python-dependencies.lock) can trim several minutes off CI.
47-48: Pin Ty’s version to avoid CI breakage
uvx ty checkassumes Ty is present and compatible. To prevent sudden failures when a new Ty release lands, declare an explicit version range in the dev-dependencies section ofpyproject.toml, e.g.:-[tool.poetry.group.dev.dependencies] -ty = "*" +ty = "==0.1.*"(or the equivalent for your build tool).
📜 Review details
Configuration used: CodeRabbit UI
Review profile: CHILL
Plan: Pro
📒 Files selected for processing (1)
.github/workflows/format_and_lint.yml(1 hunks)
🧰 Additional context used
🧠 Learnings (1)
.github/workflows/format_and_lint.yml (1)
Learnt from: CR
PR: Kiln-AI/Kiln#0
File: .cursor/rules/project.mdc:0-0
Timestamp: 2025-07-21T15:28:29.450Z
Learning: To run backend / Python tests, run uv run python3 -m pytest . -s -v or a variant of it to run individual tests
🔇 Additional comments (2)
.github/workflows/format_and_lint.yml (2)
39-42: Lint step only checks import sorting – verify this is deliberate
uvx ruff check --select I .restricts diagnostics to the “I” (isort) rule set. If the intention is full lint coverage, drop--select Iand rely on the project’sruff.tomlto drive rule selection.
43-46: Formatting gate withruff format --checklooks good
The new step enforces code formatting in CI without mutating the tree – nice addition.
…ur of ty! Much less memory. Update contribting.md and .vscode/extensions.json to match
Both sounds good! integrated |
|
@leonardmq made those changes, please review when you have a second. Thoughts on merging this vs waiting for ty to hit beta? |
There was a problem hiding this comment.
Actionable comments posted: 2
🧹 Nitpick comments (1)
.vscode/extensions.json (1)
1-22: Minor style nit – consistent indentationThis file mixes tabs and spaces; VS Code’s default “JSON with Comments” formatter will normalise to spaces.
Run Format-on-Save once to avoid noisy diffs later.
📜 Review details
Configuration used: CodeRabbit UI
Review profile: CHILL
Plan: Pro
📒 Files selected for processing (6)
.vscode/extensions.json(1 hunks).vscode/settings.json(1 hunks)CONTRIBUTING.md(1 hunks)app/desktop/studio_server/finetune_api.py(5 hunks)app/desktop/studio_server/provider_api.py(5 hunks)pyrightconfig.json(0 hunks)
💤 Files with no reviewable changes (1)
- pyrightconfig.json
🚧 Files skipped from review as they are similar to previous changes (3)
- CONTRIBUTING.md
- app/desktop/studio_server/provider_api.py
- app/desktop/studio_server/finetune_api.py
🧰 Additional context used
🪛 Biome (2.1.2)
.vscode/extensions.json
[error] 15-15: Expected an array, an object, or a literal but instead found ']'.
Expected an array, an object, or a literal here.
(parse)
[error] 21-21: Expected an array, an object, or a literal but instead found ']'.
Expected an array, an object, or a literal here.
(parse)
⏰ Context from checks skipped due to timeout of 90000ms. You can increase the timeout in your CodeRabbit configuration to a maximum of 15 minutes (900000ms). (5)
- GitHub Check: Build Desktop Apps (ubuntu-22.04-arm)
- GitHub Check: Build Desktop Apps (ubuntu-22.04)
- GitHub Check: Build Desktop Apps (macos-13)
- GitHub Check: Build Desktop Apps (macos-latest)
- GitHub Check: Build Desktop Apps (windows-latest)
🔇 Additional comments (1)
.vscode/settings.json (1)
18-18: Confirmpython.languageServervalue is still respected by the latest Python extension
"python.languageServer": "None"was deprecated in recent versions of the MS Python extension; the documented fallback is now"Default".
Please double-check that “None” still disables Pylance in your current toolchain; otherwise developers may silently get Pylance re-enabled.
There was a problem hiding this comment.
That it does not support typing on self (which seems basic) yet is a little concerning as we might run into similar problems later when we do not expect it.
That said, I tried this new config with a few PRs (including the "allow unicode in model names" one, which has some weird Pydantic typing), and ty seems to work well enough with what we have, so probably fine to start using it now.
In any case, easy to rollback if we ever need to - so maybe let's do it
Edit: actually found a problem where it does not flag missing args into a Pydantic object when the class is defined like this:
class PydanticObjectWithArbitraryObjectFields(BaseModel):
model_config = ConfigDict(arbitrary_types_allowed=True)
xxx: SomeObjectType = Field(
description="The rag config to use for the pipeline",
)
# does not detect the args are missing in the instantiation (pyright does catch it)
x = PydanticObjectWithArbitraryObjectFields()|
@leonardmq yeah lets wait a bit for beta, or at least |
|
awaiting astral-sh/ty#159 and beta designation. search and kill |
The folks behind uv and ruff have a type checker! It's about 3 billion times faster than pyright so let's use it.
Summary by CodeRabbit