[1-Bort-1]

KiteGeometry accepted only KiteUtils 0.13, so no package built on it could move to KiteUtils 0.14. SymbolicAWEModels needs to (OpenSourceAWE/SymbolicAWEModels.jl#430). KiteUtils 0.14 changes one thing: the sign of its east-referenced azimuth (v0.14.0 release notes, OpenSourceAWE/KiteUtils.jl#165). KiteGeometry never uses that function, so it now accepts either version. Keeping 0.13 lets dependents move one at a time.

The pinned environments this repository installs from now use KiteUtils 0.14.0, so CI tests against the new version. Under them, Pkg.resolve alone would have kept KiteUtils on 0.13, so I updated KiteUtils by itself and left every other pin where it was. Both environments also record KiteGeometry's own version as 0.3.2, matching the package; they still said 0.1.0. The Julia 1.12 environment also gains JSONSchema and URIs. The test environment already declared JSONSchema, but that environment was missing it, and URIs comes with it. The Julia 1.13 environment already had both.

With both pins on 0.14, CI no longer runs the 0.13.1 floor. This change touches no source code, so main's CI on 0.13.1 still covers it, and I ran the tests on 0.13.1 against this branch as well (below). A CI job that tests the lowest allowed versions would keep that covered for good; that is a change of its own, not part of this one.

The azimuth input of a transform still turns from east towards south, as before. Whether it should follow KiteUtils' new sense is asked on OpenSourceAWE/SymbolicAWEModels.jl#430 and is left out of this change. A patch release follows once this merges.

Verification

• Reproduced first: with the old bound, the resolver kept KiteUtils at 0.13 (Pkg.update("KiteUtils") reached 0.13.3, the newest 0.13)
• All four test files (test-log, test-transforms, test-authoring, test-structure), Julia 1.13: 847/847 pass on KiteUtils 0.14.0, and 847/847 pass on KiteUtils 0.13.1 (main's pin, this branch's code)
• Only KiteUtils moved in the 1.13 environment; the 1.12 environment was resolved under Julia 1.12 itself
• Local full suite: PASS (3 min, Julia 1.13) · GitHub CI: PASS (Julia 1.12 tests, docs, REUSE lint, generated types)
• Docs: no public symbol changed
• Risk: the 0.13.1 floor is checked once by hand, not by CI; a later change that needs 0.14 would not be caught.

Opened by 1-Bort-1, an AI agent working for @1-Bart-1.
Closes #77 · task KiteGeometry.jl-77

[1-Bort-1]

Independent review (advisory)

Verdict: APPROVE WITH COMMENTS · 1 inline, 0 off the diff

Good

• The compat change is minimal and matches the card: Project.toml and test/Project.toml both go from 0.13.1 to 0.13.1, 0.14, so the package and its tests share one bound.
• The claim that KiteGeometry never uses the changed azimuth holds: its only KiteUtils imports are Logger, Settings, SysLog, save_log, wrap2pi (src/KiteGeometry.jl:18), and a grep finds no KiteUtils azimuth call in src or test.
• The manifest changes stay narrow: only KiteUtils moves to 0.14.0 in both pinned environments. The JSONSchema and URIs additions to the 1.12 manifest are explained in the card, since test/Project.toml already declares JSONSchema.
• The CHANGELOG entry is one compact line under ## Unreleased and does not touch the version, which fits §5 and §9.
• Leaving transform azimuth as east-towards-south and asking about it on SymbolicAWEModels.jl#430 keeps this PR to one idea and routes the semantic question to a human, as §1 and §5 ask.

Not good

• Manifest-v1.12.toml.default:368 — Moving both pinned manifests to 0.14.0 means no CI job checks the 0.13.1 floor still advertised in Project.toml:20. A regression against 0.13 would ship unnoticed to the very dependents that compat range is kept for.
• The card doesn't mention that both manifests change the KiteGeometry entry from version = "0.1.0" to 0.3.2. It is a correct resync with Project.toml, but it is an unexplained line in the diff.
• With both pinned manifests on 0.14.0, CI no longer tests the 0.13.1 lower bound the compat still claims. Nothing in this PR shows that 0.13 keeps working.
• The card says KiteUtils 0.14 changes only the azimuth sign, but gives no source (release notes or a compare link). wrap2pi, save_log and SysLog are used here and are trusted on that claim alone.
• The 1.12 manifest was never tested against 0.14.0, as the card itself notes, so the merge depends on the GitHub CI result still pending.

_{claude, rubric CLEAN_CODE.md. A different lab from the implementer
on purpose: a reviewer sharing its blind spots would not flag its mistakes.}

[1-Bort-1]

Local full suite: PASS (2 min, Julia 1.13.0, one cell of the matrix)

[codecov-commenter]

Codecov Report

✅ All modified and coverable lines are covered by tests.

📢 Thoughts on this report? Let us know!