fix: don't signal onComplete when a chunk is fully consumed by the ra… - #523
Merged
Merged
Conversation
| // rather than signaling completion. | ||
| numBytesToSkip -= buf.length; | ||
| wrappedSubscriber.onComplete(); | ||
| return; |
Contributor
There was a problem hiding this comment.
I am pretty sure we still need to signal something to the subscriber. I don't know what it would be but it feels really weird that we would go from telling the subscriber "done" to not telling the subscriber anything.
Contributor
There was a problem hiding this comment.
Maybe signal wrappedSubscriber's onNext with an empty buffer? Since that's really what's going on here
There was a problem hiding this comment.
Same concern. Should we just do .onComplete() and return and wait for the next turn instead of signaling it here?
Contributor
Author
There was a problem hiding this comment.
I have updated the PR with onNext. Can you please refer to the latest diff?
…nge skip AdjustedRangeSubscriber called onComplete (and fell through to an NPE) when numBytesToSkip exceeded a chunk, so a small first chunk ended ranged GETs with an empty stream. Consume the chunk toward the skip and wait for the next one instead. Fixes #517.
sharmabikram
force-pushed
the
shbikram/adjustedrange-premature-oncomplete
branch
from
September 26, 2026 23:03
f9fb658 to
4bb8003
Compare
The existing coverage exercises ranged GETs over real transport, which only intermittently produces a first chunk smaller than the skip, so it cannot deterministically catch the premature-onComplete bug (#517). RangedGetUtilsTest covers only the range string math and never drives the subscriber. Add tests that wire the real production topology in-JVM -- a backpressure- honoring publisher feeding AdjustedRangeSubscriber wrapping the real InputStreamSubscriber (as toBlockingInputStream uses) -- and assert the full in-range payload is delivered when the first chunk is smaller than, empty, or exactly equal to the skip. Each read is bounded by a timeout so a demand stall fails fast instead of hanging. These fail on the pre-fix code (empty stream) and pass with the fix. No AWS credentials required; runs in the normal test phase.
Drop editorializing ("REAL", issue references) from the javadoc in favor of
plain descriptions of the topology and each case.
sharmabikram
marked this pull request as draft
September 30, 2026 20:52
Shorten the skip-branch comment and the test comments to one-liners, remove issue references, and drop narration that just restates the adjacent code.
sharmabikram
marked this pull request as ready for review
September 30, 2026 21:47
hanabanaka
approved these changes
Oct 1, 2026
hanabanaka
reviewed
Oct 1, 2026
| @Override | ||
| public void onNext(ByteBuffer byteBuffer) { | ||
| if (virtualAvailable <= 0) { | ||
| wrappedSubscriber.onComplete(); |
There was a problem hiding this comment.
(not blocking) We probably should also return here so that onComplete doesn't get called twice. But it's not producing the kind of problem that we want to fix.
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
4 participants
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.
Consume the chunk toward the skip and return instead of completing; the real onComplete still fires when upstream ends. Changed > to >= so a chunk equal to the skip is also fully consumed. Fixes #517.
Issue #, if available:
Description of changes:
By submitting this pull request, I confirm that my contribution is made under the terms of the Apache 2.0 license.
Check any applicable: