refactor(SendStream): remove duplicate buf.is_empty check

[mxinden]

SendStream::send_internal starts off with a check to buf.is_empty.

neqo/neqo-transport/src/send_stream.rs

Lines 1252 to 1256 in cf00098

	fn send_internal(&mut self, buf: &[u8], atomic: bool) -> Res<usize> {
	if buf.is_empty() {
	qerror!([self], "zero-length send on stream");
	return Err(Error::InvalidInput);
	}

Thus the following buf.is_empty check isn't necessary.

neqo/neqo-transport/src/send_stream.rs

Line 1272 in cf00098

let buf = if buf.is_empty() || (self.avail() == 0) {

[github-actions]

Benchmark results

Performance differences relative to cf00098.

• drain a timer quickly time: [310.73 ns 318.52 ns 325.70 ns]
  change: [-1.9067% +0.2833% +2.6733%] (p = 0.81 > 0.05)
  No change in performance detected.

• coalesce_acked_from_zero 1+1 entries
  time: [216.76 ns 217.14 ns 217.57 ns]
  change: [+11.227% +11.748% +12.344%] (p = 0.00 < 0.05)
  💔 Performance has regressed.

• coalesce_acked_from_zero 3+1 entries
  time: [251.08 ns 251.65 ns 252.24 ns]
  change: [+4.5758% +4.9743% +5.3748%] (p = 0.00 < 0.05)
  💔 Performance has regressed.

• coalesce_acked_from_zero 10+1 entries
  time: [250.52 ns 254.94 ns 263.87 ns]
  change: [+4.5980% +5.9825% +8.1424%] (p = 0.00 < 0.05)
  💔 Performance has regressed.

• coalesce_acked_from_zero 1000+1 entries
  time: [225.56 ns 225.81 ns 226.10 ns]
  change: [+2.2059% +2.9200% +3.5889%] (p = 0.00 < 0.05)
  💔 Performance has regressed.

• RxStreamOrderer::inbound_frame()
  time: [119.98 ms 120.16 ms 120.41 ms]
  change: [+0.9961% +1.2575% +1.5190%] (p = 0.00 < 0.05)
  Change within noise threshold.

• transfer/Run multiple transfers with varying seeds
  time: [118.05 ms 118.35 ms 118.66 ms]
  thrpt: [33.709 MiB/s 33.797 MiB/s 33.885 MiB/s]
  change:
  time: [-1.0341% -0.6833% -0.3145%] (p = 0.00 < 0.05)
  thrpt: [+0.3155% +0.6880% +1.0449%]
  Change within noise threshold.

• transfer/Run multiple transfers with the same seed
  time: [119.11 ms 119.28 ms 119.46 ms]
  thrpt: [33.484 MiB/s 33.535 MiB/s 33.582 MiB/s]
  change:
  time: [-0.5077% -0.3102% -0.0952%] (p = 0.00 < 0.05)
  thrpt: [+0.0953% +0.3112% +0.5103%]
  Change within noise threshold.

• 1-conn/1-100mb-resp (aka. Download)/client
  time: [1.0978 s 1.1055 s 1.1135 s]
  thrpt: [89.810 MiB/s 90.454 MiB/s 91.094 MiB/s]
  change:
  time: [-2.2865% -1.4881% -0.5824%] (p = 0.00 < 0.05)
  thrpt: [+0.5858% +1.5105% +2.3400%]
  Change within noise threshold.

• 1-conn/10_000-parallel-1b-resp (aka. RPS)/client
  time: [423.52 ms 425.65 ms 427.83 ms]
  thrpt: [23.374 Kelem/s 23.494 Kelem/s 23.612 Kelem/s]
  change:
  time: [-1.7004% -0.9396% -0.1942%] (p = 0.02 < 0.05)
  thrpt: [+0.1946% +0.9486% +1.7298%]
  Change within noise threshold.

• 1-conn/1-1b-resp (aka. HPS)/client
  time: [50.538 ms 50.812 ms 51.133 ms]
  thrpt: [19.557 elem/s 19.680 elem/s 19.787 elem/s]
  change:
  time: [-0.4236% +0.4634% +1.3910%] (p = 0.33 > 0.05)
  thrpt: [-1.3719% -0.4613% +0.4254%]
  No change in performance detected.

Client/server transfer results

Transfer of 134217728 bytes over loopback.

Client	Server	CC	Pacing	Mean [ms]	Min [ms]	Max [ms]	Relative
msquic	msquic			404.7 ± 19.2	365.7	437.6	1.00
neqo	msquic	reno	on	813.4 ± 30.4	775.9	884.1	1.00
neqo	msquic	reno		795.1 ± 27.0	764.9	857.7	1.00
neqo	msquic	cubic	on	807.6 ± 14.0	788.2	826.1	1.00
neqo	msquic	cubic		801.4 ± 21.0	773.3	836.1	1.00
msquic	neqo	reno	on	4355.8 ± 154.1	4137.4	4594.7	1.00
msquic	neqo	reno		4313.6 ± 159.2	4111.0	4569.3	1.00
msquic	neqo	cubic	on	4370.1 ± 147.9	4140.3	4563.7	1.00
msquic	neqo	cubic		4374.2 ± 152.7	4119.0	4577.2	1.00
neqo	neqo	reno	on	3890.4 ± 178.7	3626.8	4148.9	1.00
neqo	neqo	reno		3337.8 ± 190.7	3103.7	3675.9	1.00
neqo	neqo	cubic	on	4230.4 ± 153.4	3995.8	4433.4	1.00
neqo	neqo	cubic		4195.1 ± 437.7	3016.9	4529.9	1.00

⬇️ Download logs