[danielplohmann]

What this changes

Raises smda's declared capstone requirement from capstone<6 to capstone>=5.0.2,<6, with the rationale beside the existing comment in pyproject.toml, plus a CHANGELOG.md entry under ### Fixed.

StringExtractor imports CS_GRP_BRANCH_RELATIVE since #382, and capstone 4.x does not have that constant. With no lower bound declared, pip resolves capstone 4.0.2 whenever another package pins it. dumpulator 0.2.6 does (capstone~=4.0.2). pip check then reports nothing, and import smda.Disassembler raises ImportError. This came up in a fresh Ubuntu 26.04 course-host install that put smda 4.11.0, mcrit and dumpulator into one venv.

Why the floor is 5.0.2 and not 5.0.0: capstone 5.0.0.post1 and 5.0.1 import distutils, which Python 3.12 removed, so on the interpreters smda supports they fail at import.

Cost: smda no longer installs into the same environment as dumpulator 0.2.6. That conflict existed before as a runtime ImportError. pip now reports it at install time as ResolutionImpossible.

How it was verified

Python 3.14.4, branched from bff1bfd (v4.11.0):

capstone	Result
4.0.2	collection fails: ImportError: cannot import name 'CS_GRP_BRANCH_RELATIVE'
5.0.0.post1, 5.0.1	import capstone fails: No module named 'distutils'
5.0.2	full suite: 2397 passed, 2 skipped, 2789 subtests passed
5.0.9	full suite: 2397 passed, 2 skipped, 2789 subtests passed

• Installed metadata reads capstone<6,>=5.0.2.

• pip install --dry-run . dumpulator==0.2.6 now fails with ResolutionImpossible, where it used to install.

• make lint and make format (ruff check: all passed; ruff format: 304 files already formatted)

• make test-all (full suite on 5.0.2 and 5.0.9, see the table)

• CHANGELOG.md entry under ## [Unreleased]

• Golden fixtures unchanged (metadata-only change, no source touched)

🤖 Generated with Claude Code

[github-actions]

📊 SMDA Performance Evaluation Benchmark Results

Generated on: 2026-10-16 13:31:30

Summary

Metric	Result
Correctness	✅ PASS — 0 / 155 common file(s) differ
Determinism	✅ PASS — base 3 run(s), PR 3 run(s)
Verdict	inconclusive — within run-to-run noise (±1.5%)
Median paired speedup	+0.36% (95% CI [+0.17%, +0.61%])
Timing noise band	±1.5%

Per-side Timing Context (best of 3/3 runs per file)

Side	Files	Functions	Median best time/file (s)	Sum of best times (s)	Throughput estimate (func/s)
base	155	173540	0.6138	215.23	~806
pr	155	173540	0.6205	214.50	~809

These context rows sum each file's best observed time across repeated runs; they are normalized comparison estimates, not single-run CI wall-clock times.

Paired per-file timing (positive speedup = PR faster):

Statistic	Value
Files compared	155
Median paired speedup	+0.36% (95% CI [+0.17%, +0.61%])
Mean speedup	+0.06% (95% CI [-0.60%, +0.68%])
Std dev / IQR	4.08% / 3.05%
Wilcoxon signed-rank p	0.0088 (n=155)

ℹ️ base and PR are timed in the same job on the same runner, with their passes interleaved base/PR/PR/base/base/PR so that drift over the job cancels rather than landing on whichever side ran last. The noise band above is therefore run-to-run variation on one machine, which is what it is able to measure; a difference inside it is reported as inconclusive. Correctness and determinism are not timing-based and are unaffected either way.

Determinism (self-check across repeated runs)

Side	Runs	Files	Deterministic	Median timing CV
base	3	155	✅	1.5%
pr	3	155	✅	1.3%

Pairwise run matrix (individual run medians, diagnostic)

Diagnostic only: each row compares one raw PR run against one raw base run. The headline verdict above uses paired per-file best-of-runs timings.

Comparison Run	PR Files	Base Files	Common	Function Set Matches	Med PR (s)	Med Base (s)	Med Diff (s)	Speedup %
pr_0 vs base_0	155	155	155	155/155	0.6382s	0.6583s	-0.0201s	+3.06%
pr_0 vs base_1	155	155	155	155/155	0.6382s	0.6270s	+0.0112s	-1.78%
pr_0 vs base_2	155	155	155	155/155	0.6382s	0.6193s	+0.0188s	-3.04%
pr_1 vs base_0	155	155	155	155/155	0.6399s	0.6583s	-0.0183s	+2.79%
pr_1 vs base_1	155	155	155	155/155	0.6399s	0.6270s	+0.0129s	-2.06%
pr_1 vs base_2	155	155	155	155/155	0.6399s	0.6193s	+0.0206s	-3.33%
pr_2 vs base_0	155	155	155	155/155	0.6346s	0.6583s	-0.0237s	+3.60%
pr_2 vs base_1	155	155	155	155/155	0.6346s	0.6270s	+0.0076s	-1.21%
pr_2 vs base_2	155	155	155	155/155	0.6346s	0.6193s	+0.0153s	-2.46%