mxsm · mxsm · Jun 24, 2026 · Jun 24, 2026
diff --git a/Cargo.toml b/Cargo.toml
@@ -69,3 +69,8 @@ harness = false
 [[bench]]
 name = "simd"
 harness = false
+
+[[bench]]
+name = "packed"
+harness = false
+required-features = ["experimental-packed"]
diff --git a/bench-results/README.md b/bench-results/README.md
@@ -20,6 +20,15 @@ bench-results/
     topic.json
     properties.json
     remoting-header.json
+  packed-evidence/
+    YYYYMMDD-HHMMSS/
+      summary.md
+      packed-test.txt
+      packed-miri.txt
+      packed-asan.txt
+      fuzz-packed-from-bytes.txt
+      fuzz-packed-push-str.txt
+      packed-bench.txt
   summaries/
     summary-v1.1-v1.2.md
     summary-v1.2-v2-packed.md
@@ -51,3 +60,14 @@ On Windows PowerShell:
 ```powershell
 scripts/bench-all.ps1 current
 ```
+
+For the experimental packed representation evidence gate:
+
+```powershell
+scripts/verify-packed.ps1 -RunMiri -RunSanitizer -RunFuzz -RunBench
+```
+
+The packed evidence script always runs `cargo test --features experimental-packed`.
+The Miri, sanitizer, fuzz, and packed benchmark gates are opt-in because they
+require nightly components, optional cargo subcommands, target support, or more
+runtime than the default test suite.
diff --git a/bench-results/packed-evidence/20260624-130416/fuzz-packed-from-bytes.txt b/bench-results/packed-evidence/20260624-130416/fuzz-packed-from-bytes.txt
@@ -0,0 +1,25 @@
+warning: linker stdout: 姝ｅ湪鍒涘缓搴?C:\Users\ljbmx\Desktop\mxsm\cheetah-string\fuzz\target\x86_64-pc-windows-msvc\release\deps\fuzz_packed_from_bytes.lib 鍜屽璞?C:\Users\ljbmx\Desktop\mxsm\cheetah-string\fuzz\target\x86_64-pc-windows-msvc\release\deps\fuzz_packed_from_bytes.exp
+  |
+  = note: `#[warn(linker_messages)]` on by default
+
+warning: `cheetah-string-fuzz` (bin "fuzz_packed_from_bytes") generated 1 warning
+    Finished `release` profile [optimized + debuginfo] target(s) in 0.04s
+warning: linker stdout: 姝ｅ湪鍒涘缓搴?C:\Users\ljbmx\Desktop\mxsm\cheetah-string\fuzz\target\x86_64-pc-windows-msvc\release\deps\fuzz_packed_from_bytes.lib 鍜屽璞?C:\Users\ljbmx\Desktop\mxsm\cheetah-string\fuzz\target\x86_64-pc-windows-msvc\release\deps\fuzz_packed_from_bytes.exp
+  |
+  = note: `#[warn(linker_messages)]` on by default
+
+warning: `cheetah-string-fuzz` (bin "fuzz_packed_from_bytes") generated 1 warning
+    Finished `release` profile [optimized + debuginfo] target(s) in 0.01s
+     Running `fuzz\target\x86_64-pc-windows-msvc\release\fuzz_packed_from_bytes.exe -artifact_prefix=C:\Users\ljbmx\Desktop\mxsm\cheetah-string\fuzz\artifacts\fuzz_packed_from_bytes\ -runs=100 C:\Users\ljbmx\Desktop\mxsm\cheetah-string\fuzz\corpus\fuzz_packed_from_bytes`
+INFO: Running with entropic power schedule (0xFF, 100).
+INFO: Seed: 493125877
+INFO: Loaded 1 modules   (290 inline 8-bit counters): 290 [00007FF659A85EC8, 00007FF659A85FEA), 
+INFO: Loaded 1 PC tables (290 PCs): 290 [00007FF659A66BD8,00007FF659A67DF8), 
+INFO:        0 files found in C:\Users\ljbmx\Desktop\mxsm\cheetah-string\fuzz\corpus\fuzz_packed_from_bytes
+INFO: -max_len is not provided; libFuzzer will not generate inputs larger than 4096 bytes
+INFO: A corpus is not provided, starting from an empty corpus
+#2	INITED cov: 26 ft: 26 corp: 1/1b exec/s: 0 rss: 41Mb
+#4	NEW    cov: 27 ft: 27 corp: 2/3b lim: 4 exec/s: 0 rss: 41Mb L: 2/2 MS: 2 ChangeBit-InsertByte-
+#28	REDUCE cov: 27 ft: 27 corp: 2/2b lim: 4 exec/s: 0 rss: 41Mb L: 1/1 MS: 4 CrossOver-CopyPart-ChangeBinInt-EraseBytes-
+#100	DONE   cov: 27 ft: 27 corp: 2/2b lim: 4 exec/s: 0 rss: 41Mb
+Done 100 runs in 0 second(s)
diff --git a/bench-results/packed-evidence/20260624-130416/fuzz-packed-push-str.txt b/bench-results/packed-evidence/20260624-130416/fuzz-packed-push-str.txt
@@ -0,0 +1,27 @@
+warning: linker stdout: 姝ｅ湪鍒涘缓搴?C:\Users\ljbmx\Desktop\mxsm\cheetah-string\fuzz\target\x86_64-pc-windows-msvc\release\deps\fuzz_packed_push_str.lib 鍜屽璞?C:\Users\ljbmx\Desktop\mxsm\cheetah-string\fuzz\target\x86_64-pc-windows-msvc\release\deps\fuzz_packed_push_str.exp
+  |
+  = note: `#[warn(linker_messages)]` on by default
+
+warning: `cheetah-string-fuzz` (bin "fuzz_packed_push_str") generated 1 warning
+    Finished `release` profile [optimized + debuginfo] target(s) in 0.03s
+warning: linker stdout: 姝ｅ湪鍒涘缓搴?C:\Users\ljbmx\Desktop\mxsm\cheetah-string\fuzz\target\x86_64-pc-windows-msvc\release\deps\fuzz_packed_push_str.lib 鍜屽璞?C:\Users\ljbmx\Desktop\mxsm\cheetah-string\fuzz\target\x86_64-pc-windows-msvc\release\deps\fuzz_packed_push_str.exp
+  |
+  = note: `#[warn(linker_messages)]` on by default
+
+warning: `cheetah-string-fuzz` (bin "fuzz_packed_push_str") generated 1 warning
+    Finished `release` profile [optimized + debuginfo] target(s) in 0.01s
+     Running `fuzz\target\x86_64-pc-windows-msvc\release\fuzz_packed_push_str.exe -artifact_prefix=C:\Users\ljbmx\Desktop\mxsm\cheetah-string\fuzz\artifacts\fuzz_packed_push_str\ -runs=100 C:\Users\ljbmx\Desktop\mxsm\cheetah-string\fuzz\corpus\fuzz_packed_push_str`
+INFO: Running with entropic power schedule (0xFF, 100).
+INFO: Seed: 503145824
+INFO: Loaded 1 modules   (359 inline 8-bit counters): 359 [00007FF6904D7088, 00007FF6904D71EF), 
+INFO: Loaded 1 PC tables (359 PCs): 359 [00007FF6904B6BD8,00007FF6904B8248), 
+INFO:        0 files found in C:\Users\ljbmx\Desktop\mxsm\cheetah-string\fuzz\corpus\fuzz_packed_push_str
+INFO: -max_len is not provided; libFuzzer will not generate inputs larger than 4096 bytes
+INFO: A corpus is not provided, starting from an empty corpus
+#2	INITED cov: 37 ft: 37 corp: 1/1b exec/s: 0 rss: 41Mb
+#6	NEW    cov: 41 ft: 41 corp: 2/3b lim: 4 exec/s: 0 rss: 41Mb L: 2/2 MS: 4 CrossOver-ChangeByte-CrossOver-CopyPart-
+#20	NEW    cov: 42 ft: 42 corp: 3/6b lim: 4 exec/s: 0 rss: 41Mb L: 3/3 MS: 4 ChangeBit-InsertByte-CopyPart-ChangeByte-
+#31	REDUCE cov: 42 ft: 42 corp: 3/5b lim: 4 exec/s: 0 rss: 41Mb L: 2/2 MS: 1 EraseBytes-
+#34	NEW    cov: 43 ft: 43 corp: 4/7b lim: 4 exec/s: 0 rss: 41Mb L: 2/2 MS: 3 ShuffleBytes-InsertByte-ChangeByte-
+#100	DONE   cov: 43 ft: 43 corp: 4/7b lim: 4 exec/s: 0 rss: 41Mb
+Done 100 runs in 0 second(s)
diff --git a/bench-results/packed-evidence/20260624-130416/packed-asan.txt b/bench-results/packed-evidence/20260624-130416/packed-asan.txt
@@ -0,0 +1,93 @@
+
+running 1 test
+test packed::tests::packed_size_is_three_words ... ok
+
+test result: ok. 1 passed; 0 failed; 0 ignored; 0 measured; 5 filtered out; finished in 0.01s
+
+
+running 0 tests
+
+test result: ok. 0 passed; 0 failed; 0 ignored; 0 measured; 33 filtered out; finished in 0.00s
+
+
+running 0 tests
+
+test result: ok. 0 passed; 0 failed; 0 ignored; 0 measured; 41 filtered out; finished in 0.00s
+
+
+running 0 tests
+
+test result: ok. 0 passed; 0 failed; 0 ignored; 0 measured; 0 filtered out; finished in 0.00s
+
+
+running 0 tests
+
+test result: ok. 0 passed; 0 failed; 0 ignored; 0 measured; 8 filtered out; finished in 0.00s
+
+
+running 0 tests
+
+test result: ok. 0 passed; 0 failed; 0 ignored; 0 measured; 1 filtered out; finished in 0.00s
+
+
+running 0 tests
+
+test result: ok. 0 passed; 0 failed; 0 ignored; 0 measured; 8 filtered out; finished in 0.00s
+
+
+running 1 test
+test packed_layout_is_24_bytes ... ok
+
+test result: ok. 1 passed; 0 failed; 0 ignored; 0 measured; 5 filtered out; finished in 0.00s
+
+
+running 0 tests
+
+test result: ok. 0 passed; 0 failed; 0 ignored; 0 measured; 9 filtered out; finished in 0.00s
+
+
+running 0 tests
+
+test result: ok. 0 passed; 0 failed; 0 ignored; 0 measured; 5 filtered out; finished in 0.00s
+
+
+running 0 tests
+
+test result: ok. 0 passed; 0 failed; 0 ignored; 0 measured; 0 filtered out; finished in 0.00s
+
+
+running 0 tests
+
+test result: ok. 0 passed; 0 failed; 0 ignored; 0 measured; 0 filtered out; finished in 0.00s
+
+
+running 0 tests
+
+test result: ok. 0 passed; 0 failed; 0 ignored; 0 measured; 4 filtered out; finished in 0.00s
+
+
+running 0 tests
+
+test result: ok. 0 passed; 0 failed; 0 ignored; 0 measured; 22 filtered out; finished in 0.00s
+
+
+running 0 tests
+
+test result: ok. 0 passed; 0 failed; 0 ignored; 0 measured; 5 filtered out; finished in 0.00s
+
+    Finished `test` profile [unoptimized + debuginfo] target(s) in 0.15s
+     Running unittests src\lib.rs (target\debug\deps\cheetah_string-ef19ecdb4ae50af0.exe)
+     Running tests\api_extensions.rs (target\debug\deps\api_extensions-265ea9d412c4d589.exe)
+     Running tests\basic.rs (target\debug\deps\basic-5a9db4b2e7a37375.exe)
+     Running tests\bytes.rs (target\debug\deps\bytes-28bff98899307ff2.exe)
+     Running tests\comprehensive_tests.rs (target\debug\deps\comprehensive_tests-9762dbcc65ca96cf.exe)
+     Running tests\layout_snapshot.rs (target\debug\deps\layout_snapshot-178567cc15cd9278.exe)
+     Running tests\mutation.rs (target\debug\deps\mutation-261bb87b5dc079ef.exe)
+     Running tests\packed.rs (target\debug\deps\packed-4719a7e1f01d1234.exe)
+     Running tests\patterns.rs (target\debug\deps\patterns-4bb1c92062e3a356.exe)
+     Running tests\search.rs (target\debug\deps\search-70c948bc42af4b07.exe)
+     Running tests\serde.rs (target\debug\deps\serde-64c9281c5b632aae.exe)
+     Running tests\simd.rs (target\debug\deps\simd-9ebcc041882bf452.exe)
+     Running tests\split_edge_cases.rs (target\debug\deps\split_edge_cases-c696d0bfbb0ebff4.exe)
+     Running tests\sso.rs (target\debug\deps\sso-fe28e8b20a68cdd8.exe)
+     Running tests\type_split.rs (target\debug\deps\type_split-936df32e17a021ec.exe)
diff --git a/bench-results/packed-evidence/20260624-130416/packed-bench.txt b/bench-results/packed-evidence/20260624-130416/packed-bench.txt
@@ -0,0 +1,135 @@
+packed_construction/CheetahString short
+                        time:   [9.6566 ns 9.7726 ns 9.9076 ns]
+                        change: [-1.9188% +1.3898% +4.6098%] (p = 0.41 > 0.05)
+                        No change in performance detected.
+Found 13 outliers among 100 measurements (13.00%)
+  12 (12.00%) high mild
+  1 (1.00%) high severe
+packed_construction/PackedCheetahString short
+                        time:   [24.627 ns 25.266 ns 25.947 ns]
+                        change: [+8.8753% +11.638% +14.563%] (p = 0.00 < 0.05)
+                        Performance has regressed.
+Found 9 outliers among 100 measurements (9.00%)
+  4 (4.00%) high mild
+  5 (5.00%) high severe
+packed_construction/CheetahString long
+                        time:   [33.496 ns 33.947 ns 34.421 ns]
+                        change: [+2.2096% +4.7316% +7.0119%] (p = 0.00 < 0.05)
+                        Performance has regressed.
+Found 11 outliers among 100 measurements (11.00%)
+  9 (9.00%) high mild
+  2 (2.00%) high severe
+packed_construction/PackedCheetahString long
+                        time:   [21.843 ns 22.464 ns 23.123 ns]
+                        change: [+5.9260% +8.5757% +11.188%] (p = 0.00 < 0.05)
+                        Performance has regressed.
+Found 8 outliers among 100 measurements (8.00%)
+  7 (7.00%) high mild
+  1 (1.00%) high severe
+
+packed_push_str/CheetahString inline append
+                        time:   [21.574 ns 21.770 ns 21.967 ns]
+                        change: [+2.5228% +3.8924% +5.0881%] (p = 0.00 < 0.05)
+                        Performance has regressed.
+Found 1 outliers among 100 measurements (1.00%)
+  1 (1.00%) high mild
+packed_push_str/PackedCheetahString inline append
+                        time:   [12.594 ns 12.698 ns 12.819 ns]
+                        change: [+0.5722% +1.7418% +2.8416%] (p = 0.00 < 0.05)
+                        Change within noise threshold.
+packed_push_str/CheetahString promote to heap
+                        time:   [52.082 ns 52.768 ns 53.515 ns]
+                        change: [-1.7095% +0.3918% +2.4332%] (p = 0.71 > 0.05)
+                        No change in performance detected.
+Found 1 outliers among 100 measurements (1.00%)
+  1 (1.00%) high mild
+packed_push_str/PackedCheetahString promote to heap
+                        time:   [68.553 ns 70.513 ns 72.384 ns]
+                        change: [-3.8334% -1.0159% +1.7525%] (p = 0.48 > 0.05)
+                        No change in performance detected.
+Found 3 outliers among 100 measurements (3.00%)
+  3 (3.00%) high mild
+
+packed_mq_topic_insert/CheetahString short topics
+                        time:   [33.419 碌s 34.183 碌s 35.004 碌s]
+                        thrpt:  [28.568 Melem/s 29.255 Melem/s 29.923 Melem/s]
+                 change:
+                        time:   [-0.1583% +2.5345% +5.4747%] (p = 0.08 > 0.05)
+                        thrpt:  [-5.1905% -2.4719% +0.1586%]
+                        No change in performance detected.
+Found 1 outliers among 100 measurements (1.00%)
+  1 (1.00%) high mild
+packed_mq_topic_insert/PackedCheetahString short topics
+                        time:   [52.003 碌s 53.170 碌s 54.214 碌s]
+                        thrpt:  [18.445 Melem/s 18.807 Melem/s 19.230 Melem/s]
+                 change:
+                        time:   [+3.6370% +7.3587% +11.110%] (p = 0.00 < 0.05)
+                        thrpt:  [-9.9992% -6.8543% -3.5093%]
+                        Performance has regressed.
+packed_mq_topic_insert/CheetahString long topics
+                        time:   [65.165 碌s 66.119 碌s 67.130 碌s]
+                        thrpt:  [14.896 Melem/s 15.124 Melem/s 15.346 Melem/s]
+                 change:
+                        time:   [+5.4655% +7.9547% +10.517%] (p = 0.00 < 0.05)
+                        thrpt:  [-9.5165% -7.3686% -5.1822%]
+                        Performance has regressed.
+packed_mq_topic_insert/PackedCheetahString long topics
+                        time:   [50.294 碌s 51.659 碌s 52.852 碌s]
+                        thrpt:  [18.921 Melem/s 19.358 Melem/s 19.883 Melem/s]
+                 change:
+                        time:   [-16.933% -14.191% -11.417%] (p = 0.00 < 0.05)
+                        thrpt:  [+12.889% +16.538% +20.384%]
+                        Performance has improved.
+
+   Compiling cheetah-string v2.0.0 (C:\Users\ljbmx\Desktop\mxsm\cheetah-string)
+    Finished `bench` profile [optimized] target(s) in 1.72s
+     Running benches\packed.rs (target\release\deps\packed-5044d38116574fc7.exe)
+Gnuplot not found, using plotters backend
+Benchmarking packed_construction/CheetahString short
+Benchmarking packed_construction/CheetahString short: Warming up for 3.0000 s
+Benchmarking packed_construction/CheetahString short: Collecting 100 samples in estimated 5.0000 s (533M iterations)
+Benchmarking packed_construction/CheetahString short: Analyzing
+Benchmarking packed_construction/PackedCheetahString short
+Benchmarking packed_construction/PackedCheetahString short: Warming up for 3.0000 s
+Benchmarking packed_construction/PackedCheetahString short: Collecting 100 samples in estimated 5.0001 s (203M iterations)
+Benchmarking packed_construction/PackedCheetahString short: Analyzing
+Benchmarking packed_construction/CheetahString long
+Benchmarking packed_construction/CheetahString long: Warming up for 3.0000 s
+Benchmarking packed_construction/CheetahString long: Collecting 100 samples in estimated 5.0000 s (147M iterations)
+Benchmarking packed_construction/CheetahString long: Analyzing
+Benchmarking packed_construction/PackedCheetahString long
+Benchmarking packed_construction/PackedCheetahString long: Warming up for 3.0000 s
+Benchmarking packed_construction/PackedCheetahString long: Collecting 100 samples in estimated 5.0000 s (223M iterations)
+Benchmarking packed_construction/PackedCheetahString long: Analyzing
+Benchmarking packed_push_str/CheetahString inline append
+Benchmarking packed_push_str/CheetahString inline append: Warming up for 3.0000 s
+Benchmarking packed_push_str/CheetahString inline append: Collecting 100 samples in estimated 5.0001 s (233M iterations)
+Benchmarking packed_push_str/CheetahString inline append: Analyzing
+Benchmarking packed_push_str/PackedCheetahString inline append
+Benchmarking packed_push_str/PackedCheetahString inline append: Warming up for 3.0000 s
+Benchmarking packed_push_str/PackedCheetahString inline append: Collecting 100 samples in estimated 5.0000 s (393M iterations)
+Benchmarking packed_push_str/PackedCheetahString inline append: Analyzing
+Benchmarking packed_push_str/CheetahString promote to heap
+Benchmarking packed_push_str/CheetahString promote to heap: Warming up for 3.0000 s
+Benchmarking packed_push_str/CheetahString promote to heap: Collecting 100 samples in estimated 5.0002 s (93M iterations)
+Benchmarking packed_push_str/CheetahString promote to heap: Analyzing
+Benchmarking packed_push_str/PackedCheetahString promote to heap
+Benchmarking packed_push_str/PackedCheetahString promote to heap: Warming up for 3.0000 s
+Benchmarking packed_push_str/PackedCheetahString promote to heap: Collecting 100 samples in estimated 5.0000 s (76M iterations)
+Benchmarking packed_push_str/PackedCheetahString promote to heap: Analyzing
+Benchmarking packed_mq_topic_insert/CheetahString short topics
+Benchmarking packed_mq_topic_insert/CheetahString short topics: Warming up for 3.0000 s
+Benchmarking packed_mq_topic_insert/CheetahString short topics: Collecting 100 samples in estimated 5.0949 s (146k iterations)
+Benchmarking packed_mq_topic_insert/CheetahString short topics: Analyzing
+Benchmarking packed_mq_topic_insert/PackedCheetahString short topics
+Benchmarking packed_mq_topic_insert/PackedCheetahString short topics: Warming up for 3.0000 s
+Benchmarking packed_mq_topic_insert/PackedCheetahString short topics: Collecting 100 samples in estimated 5.0093 s (106k iterations)
+Benchmarking packed_mq_topic_insert/PackedCheetahString short topics: Analyzing
+Benchmarking packed_mq_topic_insert/CheetahString long topics
+Benchmarking packed_mq_topic_insert/CheetahString long topics: Warming up for 3.0000 s
+Benchmarking packed_mq_topic_insert/CheetahString long topics: Collecting 100 samples in estimated 5.1416 s (76k iterations)
+Benchmarking packed_mq_topic_insert/CheetahString long topics: Analyzing
+Benchmarking packed_mq_topic_insert/PackedCheetahString long topics
+Benchmarking packed_mq_topic_insert/PackedCheetahString long topics: Warming up for 3.0000 s
+Benchmarking packed_mq_topic_insert/PackedCheetahString long topics: Collecting 100 samples in estimated 5.1184 s (96k iterations)
+Benchmarking packed_mq_topic_insert/PackedCheetahString long topics: Analyzing