feat(core): add opt-in buffering for raw invocation event logging by ncipollina · Pull Request #398 · LayeredCraft/minimal-lambda

ncipollina · 2026-09-04T12:11:40Z

Summary

Middleware couldn't safely log the raw Lambda event payload: IInvocationDataFeature.EventStream is read once, lazily, by event deserialization, and isn't guaranteed to be seekable — reading it in middleware would starve deserialization.
Adds an opt-in EnableBuffering() on IInvocationDataFeature (same idea as ASP.NET Core's HttpRequest.EnableBuffering()): buffers the event stream into a seekable MemoryStream only when a caller asks for it, so the cost is paid only by invocations that actually log the raw payload.
Adds ILambdaInvocationContext.EnableEventBuffering() as an ergonomic wrapper, matching the existing GetEvent<T>/GetResponse<T> extension pattern.

Changes

IInvocationDataFeature.EnableBuffering() — new method; no-ops if the stream is already seekable.
InvocationDataFeature.EnableBuffering() — copies the event stream into a MemoryStream, disposes the original, resets Position to 0.
ILambdaInvocationContext.EnableEventBuffering() extension in FeatureLambdaInvocationContextExtensions.
README: documents the pattern for logging raw request/response from middleware — buffer the event stream and reset Position after reading; the response stream can always be swapped since it's serialized to after the middleware pipeline completes.

Validation

dotnet build src/MinimalLambda/MinimalLambda.csproj — success, 0 warnings.
dotnet test tests/MinimalLambda.UnitTests (net10.0) — 545/545 passed, including new coverage for EnableBuffering (seekable no-op, non-seekable copy, content/position preservation, post-read reset) and the new context extension.
dotnet jb cleanupcode (repo formatter) run over the solution.

Notes for Reviewers

Existing EventStream behavior is unchanged for consumers who never call EnableBuffering() — factory still hands back the original stream as-is, so this is additive.

🤖 Generated with Claude Code

https://claude.ai/code/session_01Qn9kagNWsCubTrnzN6NEqV

Middleware wanting to log the raw request payload previously had no safe way to read IInvocationDataFeature.EventStream: it's read once, lazily, by event deserialization, and isn't guaranteed to be seekable. Reading it directly in middleware would starve deserialization. Add EnableBuffering() (mirrors ASP.NET Core's HttpRequest.EnableBuffering) which buffers the stream into a seekable MemoryStream only when needed, plus an ILambdaInvocationContext.EnableEventBuffering() convenience extension. Opt-in per invocation to avoid the copy when nobody is logging. Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01Qn9kagNWsCubTrnzN6NEqV

chatgpt-codex-connector

💡 Codex Review

Here are some automated review suggestions for this pull request.

Reviewed commit:30094736fe

ℹ️ About Codex in GitHub

Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you

Open a pull request for review
Mark a draft as ready
Comment "@codex review".

If Codex has suggestions, it will comment; otherwise it will react with 👍.

Codex can also answer questions or update the PR. Try commenting "@codex address that feedback".

j-d-ha

Suggestion: could we consider making EnableEventBuffering() a context-level API, with the invocation-data feature keeping the implementation detail? It feels closer to the existing GetEvent<T>() / GetResponse<T>() helpers and would let middleware opt in without needing to know where the event stream is stored.

Current middleware usage:

lambda.UseMiddleware(async(context,next)=>{varinvocationData=context.Features.GetRequired<IInvocationDataFeature>();invocationData.EnableBuffering();usingvarreader=newStreamReader(invocationData.EventStream,leaveOpen:true);logger.LogInformation("Request: {Raw}",awaitreader.ReadToEndAsync());invocationData.EventStream.Position=0;awaitnext(context);});

Possible context-level usage:

lambda.UseMiddleware(async(context,next)=>{context.EnableEventBuffering();varinvocationData=context.Features.GetRequired<IInvocationDataFeature>();usingvarreader=newStreamReader(invocationData.EventStream,leaveOpen:true);logger.LogInformation("Request: {Raw}",awaitreader.ReadToEndAsync());invocationData.EventStream.Position=0;awaitnext(context);});

Not blocking—the feature-level API is coherent since it owns the stream.

Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01Qn9kagNWsCubTrnzN6NEqV

…ing' into feat/core-invocation-data-buffering

Adding EnableBuffering() directly to the public IInvocationDataFeature interface source-breaks anyone who implements it themselves (custom test hosts, hand-rolled fakes). Split it into a separate IInvocationDataBufferingFeature, probed for via the feature collection like ASP.NET Core's optional HTTP features, so IInvocationDataFeature stays untouched and existing implementations keep compiling. Addresses review feedback from chatgpt-codex-connector on PR #398. Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01Qn9kagNWsCubTrnzN6NEqV

ncipollina · 2026-09-04T12:41:42Z

@codex review — please focus on the latest changes (commit 4e42345, the IInvocationDataBufferingFeature split addressing the earlier feedback) rather than a full repo review.

chatgpt-codex-connector

💡 Codex Review

Here are some automated review suggestions for this pull request.

Reviewed commit:4e42345cb1

ℹ️ About Codex in GitHub

Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you

Open a pull request for review
Mark a draft as ready
Comment "@codex review".

If Codex has suggestions, it will comment; otherwise it will react with 👍.

Codex can also answer questions or update the PR. Try commenting "@codex address that feedback".

j-d-ha

Great work!

The prior fix registered IInvocationDataBufferingFeature under its own key in the feature collection, alongside IInvocationDataFeature. Since Features.Set<T> keys strictly by typeof(T), those became two independent slots pointing at the same instance. Middleware replacing IInvocationDataFeature (Features.Set<IInvocationDataFeature>(...)) would leave the buffering slot pointing at the stale instance, so EnableEventBuffering() would silently buffer the wrong stream while the actually-active one still gets consumed once by deserialization - the exact starvation bug this feature exists to prevent. Drop the separate registration. EnableEventBuffering() now resolves whatever IInvocationDataFeature is currently active and probes it for the capability, so there's no second slot to fall out of sync. Addresses further review feedback from chatgpt-codex-connector on PR #398. Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01Qn9kagNWsCubTrnzN6NEqV

ncipollina · 2026-09-04T12:52:20Z

@codex review — please re-review the latest changes (commit 986bc80, resolving the buffering-desync feedback from the previous pass) rather than a full repo review.

chatgpt-codex-connector · 2026-09-04T12:54:34Z

Codex Review: Didn't find any major issues. Delightful!

Reviewed commit:986bc80eb3

ℹ️ About Codex in GitHub

Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you

Open a pull request for review
Mark a draft as ready
Comment "@codex review".

If Codex has suggestions, it will comment; otherwise it will react with 👍.

Codex can also answer questions or update the PR. Try commenting "@codex address that feedback".

j-d-ha

LGTM

github-actionsBot added the type: feat New feature label Sep 4, 2026

Merge branch 'main' into feat/core-invocation-data-buffering
bb688ae

chatgpt-codex-connectorBot reviewed Sep 4, 2026
View reviewed changes

Comment threadsrc/MinimalLambda.Abstractions/Features/IInvocationDataFeature.cs Outdated

j-d-ha previously approved these changes Sep 4, 2026
View reviewed changes

ncipollinaand others added 2 commits September 4, 2026 08:26

chore(deps): update NuGet package versions
af010c0
Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01Qn9kagNWsCubTrnzN6NEqV

Merge remote-tracking branch 'origin/feat/core-invocation-data-buffer…
9f67bc9
…ing' into feat/core-invocation-data-buffering

ncipollina dismissed j-d-ha’s stale review via 9f67bc9September 4, 2026 12:27

chatgpt-codex-connectorBot reviewed Sep 4, 2026
View reviewed changes

Comment threadsrc/MinimalLambda/Runtime/LambdaHandlerComposer.cs Outdated

j-d-ha previously approved these changes Sep 4, 2026
View reviewed changes

ncipollina dismissed j-d-ha’s stale review via 986bc80September 4, 2026 12:50

j-d-ha approved these changes Sep 4, 2026
View reviewed changes

ncipollina merged commit e7a8bc0 into mainSep 4, 2026
6 checks passed

ncipollina deleted the feat/core-invocation-data-buffering branch September 4, 2026 12:57

Navigation Menu

Uh oh!

feat(core): add opt-in buffering for raw invocation event logging - #398

feat(core): add opt-in buffering for raw invocation event logging#398
ncipollina merged 6 commits into
mainfrom
feat/core-invocation-data-buffering

ncipollina commented Sep 4, 2026

Uh oh!

chatgpt-codex-connectorBot left a comment

Uh oh!

Uh oh!

j-d-ha left a comment

Uh oh!

ncipollina commented Sep 4, 2026

Uh oh!

chatgpt-codex-connectorBot left a comment

Uh oh!

Uh oh!

j-d-ha left a comment

Uh oh!

ncipollina commented Sep 4, 2026

Uh oh!

chatgpt-codex-connectorBot commented Sep 4, 2026

Uh oh!

j-d-ha left a comment

Uh oh!

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

2 participants

Uh oh!

Conversation

ncipollina commented Sep 4, 2026

Summary

Changes

Validation

Notes for Reviewers

Uh oh!

chatgpt-codex-connectorBot left a comment

Choose a reason for hiding this comment

💡 Codex Review

Uh oh!

Uh oh!

j-d-ha left a comment

Choose a reason for hiding this comment

Uh oh!

ncipollina commented Sep 4, 2026

Uh oh!

chatgpt-codex-connectorBot left a comment

Choose a reason for hiding this comment

💡 Codex Review

Uh oh!

Uh oh!

j-d-ha left a comment

Choose a reason for hiding this comment

Uh oh!

ncipollina commented Sep 4, 2026

Uh oh!

chatgpt-codex-connectorBot commented Sep 4, 2026

Uh oh!

j-d-ha left a comment

Choose a reason for hiding this comment

Uh oh!

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

2 participants