Repository navigation
Deliver source errors through parallel SubscribeMany and the specialised aggregators - #51
Merged
Merged
Conversation
Port of upstream reactivemarbles#1208 (4769454, open), cited rather than cherry-picked because the fork's file differs. The ParallelisationOptions overloads of cache SubscribeMany subscribed their internal Transform(...).DisposeMany() pipeline with a bare Subscribe(). That subscription is made before the subscriber's, so a source error was rethrown on the producer's thread and the subscriber never received OnError. It now ignores terminal events, as DD-027 does for the internal DisposeMany subscriptions in GroupOn and TreeBuilder, and the subscriber's own subscription carries them. ParallelisedSubscribeManyFailsWhenTheSourceFails, translated to TUnit, fails on main in both flavours and passes with the change. Fixes #47 Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Port of upstream reactivemarbles#1209 (8b1a975, open), cited rather than cherry-picked because the fork's files differ. The sorted, paged and distinct changeset aggregators recorded the error in one subscription but also subscribed the messages and the CollectUpdateStats() summary with only an OnNext handler. A source error was rethrown out of those to whatever raised it, so asserting on Error got an exception instead. They now match ChangeSetAggregator and the virtual aggregator: one subscription records messages and the error, and the summary ignores terminal events. Public API unchanged. AsAggregatorFixture (new, TUnit) covers all four aggregators. The sorted, paged and distinct cases fail on main in both flavours and pass with the change; the virtual case already passed. Fixes #48 Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Add DD-047 (parallel SubscribeMany, reactivemarbles#1208) and DD-048 (specialised AsAggregator helpers, reactivemarbles#1209) as ports ahead of a sync, and map both upstream PRs to them under Adopted in the upstream summary. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
This file contains hidden or 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
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.
Fixes #47
Fixes #48
Ports ahead of a sync (docs/maintenance.md, "Urgent fixes"). Both upstream PRs are open, so each is cited in an adapted commit rather than cherry-picked, with its regression translated to TUnit.
Commits
47694543):PSubscribeManysubscribed its internalTransform(...).DisposeMany()with a bareSubscribe(), so a source error was rethrown to the producer and never reached the subscriber. It now ignores terminal events, as DD-027 does forGroupOn/TreeBuilder. Regression:SubscribeManyFixture.ParallelisedSubscribeManyFailsWhenTheSourceFails. This is a behaviour change: the error arrives asOnErrorinstead of being thrown.8b1a975d): the sorted, paged and distinctAsAggregatorhelpers now use one subscription for messages and the error, and their summary ignores terminal events, matchingChangeSetAggregatorand the virtual aggregator. Regression: newCache/AsAggregatorFixture(4 tests). Public API unchanged.docs/fork-differences.md("ported ahead of a sync; compare at the next sync"), plus Adopted rows indocs/upstream/README.md.I left the experimental
Watcher's bareSubscribe()alone: its source is an internal cache that never errors.Validation
mainin both flavours: 3 aggregator cases (virtual already passed) and the parallelSubscribeManycase. They pass with the change.global.jsonoverride, see Pin SDK 10.0.403 and install it into a job-owned root #50):dotnet packsucceeds.Overlap with #49
docs/upstream/README.mdis also changed by #49: it moves #47/#48 into "Open gaps", and this PR adds Adopted rows. After #49 merges I'll mergemainin and remove the two Open-gaps rows.docs/fork-differences.mdchanges in both, but in different rows (DD-030 vs new DD-047/048 at the end).🤖 Generated with Claude Code