chore(release): 0.11.8

[client-software-ci]

0.11.8 (2026-09-03)

Features

• model: Expose chunks_task_count_override on the v1 iterator (feat(model): Expose chunks_task_count_override on the v1 iterator #344) (085609a)
• model: Expose chunks_task_count_override on the v1 iterator (085609a)

Bug Fixes

• Keep the trailing zeros of a floatstring range element (fix: Keep the trailing zeros of a floatstring range element #345) (ffd628b)
• Keep the trailing zeros of a floatstring range element (ffd628b)
• trim a floatstring range element and unsign a zero (ffd628b)
• decide a zero spelling from the text, not the parse (ffd628b)
• Enforce the 512-character cap on a let binding identifier (fix: Enforce the 512-character cap on a let binding identifier #348) (262d3d2)
• Enforce the 512-character cap on a let binding identifier (262d3d2)
• Name the over-long binding in the let identifier error (262d3d2)

[seant-aws]

theres some repeats here could we fix

[leongdl]

Fixing

[github-actions]

Every commit in this release is listed twice (once with the PR number, once without), and the two multi-commit PRs are listed once per body bullet:

• 085609a — 2 entries (feat(model): Expose chunks_task_count_override on the v1 iterator #344 subject + 1 bullet)
• ffd628b — 4 entries (fix: Keep the trailing zeros of a floatstring range element #345 subject + 3 bullets)
• 262d3d2 — 3 entries (fix: Enforce the 512-character cap on a let binding identifier #348 subject + 2 bullets)

The counts line up exactly with the squash commit messages. Each of these is a GitHub squash merge whose body retains the original * fix: ... / * feat(model): ... lines, e.g. 085609a:

feat(model): Expose chunks_task_count_override on the v1 iterator (#344)

* feat(model): Expose chunks_task_count_override on the v1 iterator

python-semantic-release 10.x (pinned == 10.6.* in requirements-release.txt) parses squash commits by default, so it yields one release element for the subject and one for each conventional-commit bullet in the body. .semantic_release/CHANGELOG.md.j2 then renders every element, producing the duplicates.

Worth noting this is cosmetic-only for the version number here — the duplicates are all feat/fix, and with minor_tags = [] plus feat/fix both in patch_tags, 0.11.7 -> 0.11.8 is the correct bump either way. But it will recur on every release that squash-merges a multi-commit PR.

Suggest disabling squash parsing in pyproject.toml so only the squash subject is used:

[tool.semantic_release.commit_parser_options]
parse_squash_commits = false

Then regenerate this changelog section. Alternatively, if the per-bullet detail is wanted, dedupe on commit.short_hash in the template — but that would drop the distinct bullet summaries, so the config change is likely what you want.