fix: Refactor error handling and improve documentation

[askpt]

Signed-off-by: André Silva 2493377+askpt@users.noreply.github.com

This PR

This pull request includes updates to the README.md, changes to error handling in OpenFeatureClient, and improvements to the InMemoryProvider and its related tests.

Documentation updates:

• Added notes and examples to the README.md to improve clarity on logging, transaction context propagators, dependency injection, and custom providers. [1] [2] [3] [4] [5] [6] [7] [8] [9] [10] [11]

Error handling improvements:

• Removed redundant error logging methods FlagEvaluationError and FlagEvaluationErrorWithDescription in OpenFeatureClient. [1] [2] [3]

InMemoryProvider enhancements:

• Changed InMemoryProvider to return ResolutionDetails with appropriate error types instead of throwing exceptions for missing flags or type mismatches. [1] [2]

Test updates:

• Updated tests in OpenFeatureClientTests and InMemoryProviderTests to reflect changes in error handling and ensure proper error types and reasons are returned. [1] [2] [3]

Configuration changes:

• Modified global.json to update the SDK roll-forward policy from latestMajor to latestFeature.

Related Issues

Fixes #416

Notes

After evaluating the global.json, I noticed the rollForward was way too high for what we should be using. I reduced to latestFeature which is safer and would allow us to change the GitHub Actions to use the version from global.json instead of from declaring it in the action itself.

[codecov]

Codecov Report

✅ All modified and coverable lines are covered by tests.
✅ Project coverage is 85.69%. Comparing base (3c7dca3) to head (e603ffb).
⚠️ Report is 239 commits behind head on main.

Additional details and impacted files

@@            Coverage Diff             @@
##             main     #417      +/-   ##
==========================================
- Coverage   85.71%   85.69%   -0.02%     
==========================================
  Files          39       39              
  Lines        1603     1601       -2     
  Branches      171      171              
==========================================
- Hits         1374     1372       -2     
  Misses        191      191              
  Partials       38       38              

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

🚀 New features to boost your workflow:

• ❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.

[Copilot]

Copilot reviewed 5 out of 5 changed files in this pull request and generated no comments.

Comments suppressed due to low confidence (1)

test/OpenFeature.Tests/OpenFeatureClientTests.cs:187

• [nitpick] Consider adding a comment in the test to clarify that error logging is intentionally disabled during flag evaluation to help future maintainers understand this behavior.

mockedLogger.Received(0).IsEnabled(LogLevel.Error);

[Copilot]

Copilot reviewed 5 out of 5 changed files in this pull request and generated no comments.

Comments suppressed due to low confidence (1)

src/OpenFeature/Providers/Memory/InMemoryProvider.cs:115

• [nitpick] Consider returning a ResolutionDetails with an appropriate error type for type mismatches instead of throwing an exception, to align the error handling behavior with that of missing flags. If the current behavior is by design, please add documentation explaining the rationale.

throw new TypeMismatchException("flag " + flagKey + " is not of type " + typeof(T));

[kriscoleman]

changes look good to me

I agree with the change on the flag-not-found. I prefer not to use exceptions to control logic flow; that's an old anti-pattern, and those stack traces have a hidden cost.

Thanks!