Bump fulmine to 5.5.2

[nigrosimone]

Pin bump for the fulmine entry, from ^5.5.0 to ^5.5.2 (changelog), plus three changes to the entry.

In the release, measured here against the previous revision:

• a use with no path ran a regex on every hop for a mount prefix that was always 0. Removing it is -16.8% CPU per request on a chain of 100 middlewares, 23.1k to 27.6k req/s
• the ETag hashed a copy of the body instead of the body: -7.7% on a 35 byte body, -41% on 54 KB
• the generic handler drops a promise pair nobody awaited, and a response no longer allocates its chunk queue or its flush closure unless it streams

In the entry:

• /static is served on the TLS listener as well, so it subscribes to static-tls, same handler as the plaintext one.
• a stock Caddy edge in proxy/, so it subscribes to gateway-64 and gateway-h3. Caddy terminates TLS and serves /static/* from disk, the server keeps the three dynamic endpoints it already had, reached over loopback h1. I ran both stacks locally against the repo certs: h2 and h3 both negotiate, /static comes from the proxy with the right content types and a 404 for a missing file, and /baseline2, /json/N and /async-db all answer through it.
• json-comp prefers brotli, which it never reached before because Accept-Encoding: gzip, br matched gzip first. On these payloads q3 is 12% smaller than gzip level 1 for about 24us more per request, and with the score squaring the body size that should come out ahead, but it is a model and not your rig, so the run decides.

/benchmark -f fulmine --save

[nigrosimone]

benchmark -f fulmine --save

[MDA2AV]

/benchmark -f fulmine --save

[github-actions]

👋 Benchmark request received. A collaborator will review and approve the run.

[github-actions]

Benchmark Results

Framework: fulmine | Test: all tests

Test	Conn	RPS	CPU	Mem	Δ RPS	Δ Mem
baseline	512	1,335,278	6794.8%	2.6GiB	+8.5%	~0%
baseline	4096	1,297,447	6797.5%	2.6GiB	+7.0%	~0%
pipelined	512	7,917,830	6580.3%	2.5GiB	+0.3%	~0%
pipelined	4096	7,868,038	6575.8%	2.6GiB	~0%	~0%
limited-conn	512	1,096,830	6156.2%	2.6GiB	+6.3%	~0%
limited-conn	4096	1,175,660	6459.1%	2.7GiB	+6.6%	~0%
json	4096	1,141,084	6444.0%	2.6GiB	~0%	~0%
json-comp	512	341,669	6164.5%	8.0GiB	-17.1%	-8.0%
json-comp	4096	356,319	6407.9%	9.0GiB	-16.2%	-2.2%
json-comp	16384	374,554	6309.4%	10.3GiB	-14.5%	-2.8%
json-tls	4096	1,076,676	6585.5%	2.6GiB	~0%	~0%
upload	32	2,037	3081.7%	6.6GiB	+0.6%	~0%
upload	256	2,161	6256.9%	7.5GiB	+1.5%	+5.6%
api-4	256	36,387	394.5%	402MiB	+2.1%	-5.4%
api-16	1024	131,175	1540.6%	1.3GiB	+1.2%	+8.3%
static	1024	540,509	6505.3%	3.9GiB	-0.6%	~0%
static	4096	582,701	6498.8%	7.5GiB	-0.9%	+2.7%
static	6800	567,379	6558.9%	7.6GiB	~0%	~0%
static-tls	1024	432,837	6513.3%	3.7GiB	NEW	NEW
static-tls	4096	463,875	6457.0%	7.2GiB	NEW	NEW
static-tls	6800	452,587	6380.7%	7.6GiB	NEW	NEW
async-db	1024	224,608	5944.0%	3.6GiB	+0.1%	~0%
crud	4096	355,740	4401.2%	8.1GiB	-0.5%	+1.2%
gateway-64	512	88,935	4318.6%	4.5GiB	NEW	NEW
gateway-64	1024	82,340	4268.4%	6.7GiB	NEW	NEW
gateway-h3	64	91,075	3590.5%	2.6GiB	NEW	NEW
gateway-h3	256	62,791	2626.4%	3.7GiB	NEW	NEW
echo-ws	512	3,705,673	6522.3%	2.3GiB	-0.6%	~0%
echo-ws	4096	3,785,984	6549.7%	2.4GiB	+0.5%	~0%
echo-ws	16384	3,565,343	6209.8%	2.5GiB	-0.3%	~0%
echo-ws-pipeline	512	24,580,339	6740.0%	2.3GiB	-1.5%	~0%
echo-ws-pipeline	4096	24,778,835	6573.8%	2.4GiB	-0.6%	~0%
echo-ws-pipeline	16384	23,533,654	6309.9%	2.6GiB	-1.4%	~0%
echo-ws-limited	512	2,083,680	5585.6%	2.4GiB	-0.3%	+4.3%
echo-ws-limited	4096	2,557,031	6232.9%	2.4GiB	-0.3%	~0%

Full log

  Bandwidth:  153.65MB/s
  WS upgrades: 16384
  WS frames:   114720176
  Latency samples: 114720176 / 114720176 responses (100.0%)
[info] CPU 6771.9% | Mem 2.6GiB

[run 3/3]
gcannon v0.5.3 [WS]
  Target:    localhost:8080/ws
  Threads:   64
  Conns:     16384 (256/thread)
  Pipeline:  16
  Req/conn:  unlimited (keep-alive)
  Expected:  200
  Duration:  5s


  Thread Stats   Avg      p50      p90      p99    p99.9
    Latency   11.38ms   10.90ms   13.30ms   17.30ms   21.30ms

  114385824 frames sent in 5.00s, 114127776 frames received
  Throughput: 22.82M req/s
  Bandwidth:  152.85MB/s
  WS upgrades: 16384
  WS frames:   114127776
  Latency samples: 114127776 / 114127776 responses (100.0%)
[info] CPU 6489.4% | Mem 2.6GiB

=== Best: 23533654 req/s (CPU: 6309.9%, Mem: 2.6GiB) ===
[info] saved results/echo-ws-pipeline/16384/fulmine.json
httparena-bench-fulmine
httparena-bench-fulmine

==============================================
=== fulmine / echo-ws-limited / 512c (tool=gcannon) ===
==============================================
[info] waiting for server...
[info] server ready

[run 1/3]
gcannon v0.5.3 [WS]
  Target:    localhost:8080/ws
  Threads:   64
  Conns:     512 (8/thread)
  Pipeline:  1
  Req/conn:  10
  Expected:  200
  Duration:  5s


  Thread Stats   Avg      p50      p90      p99    p99.9
    Latency    145us    118us    256us    526us   1.53ms

  10311760 frames sent in 5.00s, 10311870 frames received
  Throughput: 2.06M req/s
  Bandwidth:  46.42MB/s
  WS upgrades: 1045666
  WS frames:   10311870
  Latency samples: 10311820 / 10311870 responses (100.0%)
  Reconnects: 1031237
[info] CPU 5485.4% | Mem 2.4GiB

[run 2/3]
gcannon v0.5.3 [WS]
  Target:    localhost:8080/ws
  Threads:   64
  Conns:     512 (8/thread)
  Pipeline:  1
  Req/conn:  10
  Expected:  200
  Duration:  5s


  Thread Stats   Avg      p50      p90      p99    p99.9
    Latency    146us    120us    259us    532us   1.53ms

  10418325 frames sent in 5.00s, 10418400 frames received
  Throughput: 2.08M req/s
  Bandwidth:  46.89MB/s
  WS upgrades: 1057438
  WS frames:   10418400
  Latency samples: 10420195 / 10418400 responses (100.0%)
  Reconnects: 1041876
[info] CPU 5585.6% | Mem 2.4GiB

[run 3/3]
gcannon v0.5.3 [WS]
  Target:    localhost:8080/ws
  Threads:   64
  Conns:     512 (8/thread)
  Pipeline:  1
  Req/conn:  10
  Expected:  200
  Duration:  5s


  Thread Stats   Avg      p50      p90      p99    p99.9
    Latency    146us    120us    257us    530us   1.44ms

  10370492 frames sent in 5.00s, 10370575 frames received
  Throughput: 2.07M req/s
  Bandwidth:  46.67MB/s
  WS upgrades: 1051450
  WS frames:   10370575
  Latency samples: 10370544 / 10370575 responses (100.0%)
  Reconnects: 1037105
[info] CPU 5485.7% | Mem 2.4GiB

=== Best: 2083680 req/s (CPU: 5585.6%, Mem: 2.4GiB) ===
[info] saved results/echo-ws-limited/512/fulmine.json
httparena-bench-fulmine
httparena-bench-fulmine

==============================================
=== fulmine / echo-ws-limited / 4096c (tool=gcannon) ===
==============================================
[info] waiting for server...
[info] server ready

[run 1/3]
gcannon v0.5.3 [WS]
  Target:    localhost:8080/ws
  Threads:   64
  Conns:     4096 (64/thread)
  Pipeline:  1
  Req/conn:  10
  Expected:  200
  Duration:  5s


  Thread Stats   Avg      p50      p90      p99    p99.9
    Latency   1.06ms    867us   1.75ms   3.22ms   6.13ms

  12763348 frames sent in 5.00s, 12762161 frames received
  Throughput: 2.55M req/s
  Bandwidth:  57.47MB/s
  WS upgrades: 1281777
  WS frames:   12762161
  Latency samples: 12762133 / 12762161 responses (100.0%)
  Reconnects: 1275611
[info] CPU 5961.9% | Mem 2.4GiB

[run 2/3]
gcannon v0.5.3 [WS]
  Target:    localhost:8080/ws
  Threads:   64
  Conns:     4096 (64/thread)
  Pipeline:  1
  Req/conn:  10
  Expected:  200
  Duration:  5s


  Thread Stats   Avg      p50      p90      p99    p99.9
    Latency   1.07ms    964us   1.78ms   2.87ms   3.92ms

  12785656 frames sent in 5.00s, 12785156 frames received
  Throughput: 2.56M req/s
  Bandwidth:  57.56MB/s
  WS upgrades: 1281829
  WS frames:   12785156
  Latency samples: 12785039 / 12785156 responses (100.0%)
  Reconnects: 1278194
[info] CPU 6232.9% | Mem 2.4GiB

[run 3/3]
gcannon v0.5.3 [WS]
  Target:    localhost:8080/ws
  Threads:   64
  Conns:     4096 (64/thread)
  Pipeline:  1
  Req/conn:  10
  Expected:  200
  Duration:  5s


  Thread Stats   Avg      p50      p90      p99    p99.9
    Latency   1.06ms    940us   1.76ms   2.91ms   4.48ms

  12767481 frames sent in 5.00s, 12767728 frames received
  Throughput: 2.55M req/s
  Bandwidth:  57.47MB/s
  WS upgrades: 1282544
  WS frames:   12767728
  Latency samples: 12767661 / 12767728 responses (100.0%)
  Reconnects: 1276853
[info] CPU 6053.6% | Mem 2.4GiB

=== Best: 2557031 req/s (CPU: 6232.9%, Mem: 2.4GiB) ===
[info] saved results/echo-ws-limited/4096/fulmine.json
httparena-bench-fulmine
httparena-bench-fulmine
[info] rebuilding site/data/*.json
[updated] /home/diogo/actions-runner/_work/HttpArena/HttpArena/site/data/frameworks.json
[updated] /home/diogo/actions-runner/_work/HttpArena/HttpArena/site/data/results/fulmine.json - 35 new, 35 total
[updated] /home/diogo/actions-runner/_work/HttpArena/HttpArena/site/data/current.json
[info] done
httparena-postgres
httparena-redis
[info] restoring loopback MTU to 65536

[nigrosimone]

The json-comp row reads -16% on RPS, which is the brotli change and is expected rather than a regression: that profile scores rps * (minBpr/myBpr)^2, so the bytes count twice over.

Bytes per response went 1562 to 1374, so the size term is 1.29 against 0.84 on rps. Against the field's current minBpr of 1107 the profile score goes 16.6 to 18.6.

Still well short of the 1107 the leader ships, so there is more left there, but brotli above q3 gets very slow on payloads this small and stopped paying for itself.