[rack] Parse query string with URI.decode_www_form

[p8]

URI.decode_www_form performs better than Rack::Utils which also supports parsing nested params:

require 'benchmark/ips'
require 'cgi'
require 'uri'
require 'rack'

Benchmark.ips do |x|
  x.config(warmup: 2, time: 5)

  query_string = "?a=14&b=54"

  x.report("CGI") do
    ::CGI.parse(query_string)
  end

  x.report("URI.decode_www_form") do
    URI.decode_www_form(query_string).to_h
  end

  x.report("Rack::Utils.parse_query") do
    ::Rack::Utils.parse_query(query_string)
  end

  x.compare!
end

ruby 3.4.9 (2026-03-11 revision 76cca827ab) +PRISM [arm64-darwin24]
Warming up --------------------------------------
                    CGI    60.058k i/100ms
    URI.decode_www_form    63.568k i/100ms
Rack::Utils.parse_query    40.922k i/100ms
Calculating -------------------------------------
                    CGI    632.535k (± 2.4%) i/s    (1.58 μs/i) -      3.183M in   5.032252s
    URI.decode_www_form    667.942k (± 2.0%) i/s    (1.50 μs/i) -      3.369M in   5.044006s
Rack::Utils.parse_query    417.398k (± 2.5%) i/s    (2.40 μs/i) -      2.087M in   5.000076s

Comparison:
    URI.decode_www_form:   667942.1 i/s
                    CGI:   632534.7 i/s - 1.06x  slower
Rack::Utils.parse_query:   417398.1 i/s - 1.60x  slower

[p8]

/benchmark-multiple --save

[github-actions]

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

[github-actions]

Benchmark Results

Frameworks: 2 | Test: all tests

✅ rack-falcon

Test	Conn	RPS	Rate	CPU/req	p99	p99.9	CPU	Mem	Δ RPS	Δ Mem
baseline	4096	541,466	—	—	—	—	6562.8%	1.1GiB	+3.4%	-15.4%
pipelined	4096	801,850	—	—	—	—	6666.4%	1.0GiB	+0.7%	~0%
limited-conn	4096	503,900	—	—	—	—	6255.5%	894MiB	+4.6%	-32.8%
latency-1m	1024	569,602	0.5696 ⚠️	111.96us	10940416.0us	12251136.0us	6544.5%	831MiB	+3.7%	-26.2%
latency-10k	1024	9,982	0.9982	122.50us	524.0us	12580.0us	118.8%	765MiB	~0%	-32.1%

✅ rack-iodine

Test	Conn	RPS	Rate	CPU/req	p99	p99.9	CPU	Mem	Δ RPS	Δ Mem
baseline	4096	62,831	—	—	—	—	4829.1%	4.9GiB	+0.3%	-2.0%
pipelined	4096	2,329,452	—	—	—	—	6778.4%	4.6GiB	-2.2%	~0%
limited-conn	4096	59,158	—	—	—	—	4812.8%	4.7GiB	-0.1%	-9.6%
latency-1m	1024	995,562	0.9956	56.06us	837.0us	51472.0us	6083.9%	4.4GiB	~0%	-8.3%
latency-10k	1024	9,951	0.9951	72.26us	140.0us	25256.0us	95.5%	4.3GiB	-0.2%	-10.4%