Closed achingbrain closed 7 months ago
Attention: 40 lines
in your changes are missing coverage. Please review.
Files | Coverage Δ | |
---|---|---|
src/encode.ts | 100.00% <100.00%> (ø) |
|
test/codec.spec.ts | 100.00% <100.00%> (ø) |
|
test/compliance.spec.ts | 100.00% <100.00%> (ø) |
|
test/decode.spec.ts | 96.01% <100.00%> (-0.02%) |
:arrow_down: |
test/muxer.spec.ts | 100.00% <100.00%> (ø) |
|
test/stream.spec.ts | 100.00% <100.00%> (+0.90%) |
:arrow_up: |
test/util.ts | 100.00% <100.00%> (ø) |
|
src/decode.ts | 97.22% <88.88%> (+2.08%) |
:arrow_up: |
src/stream.ts | 98.01% <99.08%> (+2.52%) |
:arrow_up: |
test/codec.util.ts | 94.28% <66.66%> (ø) |
|
... and 2 more |
:loudspeaker: Thoughts on this report? Let us know!
:tada: This PR is included in version 5.0.2 :tada:
The release is available on:
Your semantic-release bot :package::rocket:
We use an abortable source for every connection - this introduces a lot of latency.
This PR removes abortable iterator to just listen on the abort event on the shutdown controller's abort signal and cause the connection's iterator to return.
This is the same thing that abortable source does except we don't
Promise.race
every.next()
from the source.I've added a simple benchmark - in my testing it increases throughput by 30% or so, this seems to translate to 10-15% in "real world" performance using the
@libp2p/perf
testing protocol.