fix(scoop-export|scoop-list): Make exported JSON key order deterministic

[shauneccles]

Description

Two one-line changes, @{} → [ordered]@{}, so that scoop export produces byte-identical output when nothing has changed.

[PSCustomObject]$hashtable and ConvertToPrettyJson both follow the hashtable's enumeration order. For a hashtable literal PowerShell preserves the written order, so [PSCustomObject]@{ a = 1; b = 2 } is stable — but a hashtable built by assigning keys one at a time enumerates in bucket order, which depends on String.GetHashCode(), and .NET randomizes string hashing per process. Both export sites use the assignment pattern, so the key order was stable within a process and different in the next one.

• libexec/scoop-export.ps1:8 — $export is built by assigning .config / .buckets / .apps, so the three top-level keys were emitted in a random order.
• libexec/scoop-list.ps1:29 — $item is built by assigning .Name / .Version / .Source / .Updated / .Info before the [PSCustomObject] cast at line 63, so every app object in the file was independently reordered. This is what dominates the diff.

[ordered]@{} gives insertion order, which keeps the existing readable Name, Version, Source, Updated, Info layout — no sorting pass needed, and no behaviour change for consumers, since scoop import reads by property name ($item.Name, $item.Source, …).

Motivation and Context

Relates to #5862

scoop export's main use is checking the file into version control, and non-deterministic key order defeats that. On a machine committing a daily scoop export, 62 of 80 commits were pure property reordering with no content change at all — typically over 100 changed lines each, from only ~18 installed apps. Real version bumps were buried in the noise.

Worth noting #5862 is currently labelled enhancement, but the output is non-deterministic rather than merely unsorted, which is arguably a bug. Both facets reported there — the top-level order in the original report and the per-app property order in the follow-up comment — have the same root cause and are both fixed here.

How Has This Been Tested?

PowerShell 7.6.1 on Windows 11, against a real install with 18 apps and 1 bucket.

1. Determinism, running bin/scoop.ps1 export in 4 separate processes and hashing the output — once on unmodified develop, once with this change:

	distinct outputs / 4	first app's property order
develop	4	Updated, Version, Name, Info, Source
this PR	1	Name, Version, Source, Updated, Info

2. Minimal repro of the underlying mechanism (5 separate processes):

1..5 | ForEach-Object { pwsh -NoProfile -Command '$e=@{}; $e.buckets=1; $e.apps=2; $e.config=3; ($e.Keys) -join ", "' }
# apps, config, buckets / config, apps, buckets / buckets, apps, config / buckets, apps, config / config, buckets, apps

With [ordered]@{} the same loop returns insertion order every time. This is also why exporting twice in one session looks stable — the reordering only shows up across processes.

3. scoop export -c — the riskiest part of the change, since it does $export.config.PSObject.Properties.Remove(...). Verified on both variants that the output is still valid JSON, that config is present, and that last_update / root_path / global_path / cache_path / alias are still stripped. Property access and .Remove() work the same on an OrderedDictionary.

4. scoop list — unchanged output (scoop-list.ps1 backs the display path too): 22 lines, exit 0, identical on both variants.

5. Linting — Invoke-ScriptAnalyzer with the repo's PSScriptAnalyzerSettings.psd1: 0 findings on both changed files and 0 across libexec as a whole, matching Scoop-00Linting.Tests.ps1.

There is currently no test coverage for scoop export or scoop-list.ps1 in test/, so nothing existing needed updating. A regression test would have to spawn a child process to defeat per-process hash seeding, which felt disproportionate for a two-line change — happy to add one if you'd prefer.

Checklist:

• I have read the Contributing Guide.
• I have ensured that I am targeting the develop branch.
• I have updated the documentation accordingly.
• I have updated the tests accordingly.
• I have added an entry in the CHANGELOG.

[coderabbitai]

No actionable comments were generated in the recent review. 🎉

ℹ️ Recent review info

⚙️ Run configuration

Configuration used: Organization UI

Review profile: CHILL

Plan: Pro Plus

Run ID: 6d1d3306-79d4-4737-88b3-25fe5f1cd942

📥 Commits

Reviewing files that changed from the base of the PR and between 78f7221 and a819595.

📒 Files selected for processing (1)

• CHANGELOG.md

---

Summary by CodeRabbit

• Bug Fixes
  • Exported JSON now preserves deterministic key order, making output more readable and predictable.
  • App listings now use a stable field order when rendering each app’s details, without altering which apps appear.
  • Updated the changelog entry under Unreleased → Bug Fixes to reflect these output-order improvements.

Walkthrough

Changes

Output ordering

Layer / File(s)	Summary
Preserve output field order libexec/scoop-export.ps1, libexec/scoop-list.ps1, CHANGELOG.md	The export JSON payload and per-app listing records now use ordered hashtables, preserving insertion order in generated output; the changelog records this fix.

🚥 Pre-merge checks | ✅ 5

✅ Passed checks (5 passed)

Check name	Status	Explanation
Title check	✅ Passed	The title clearly matches the main change: making scoop export/list JSON key order deterministic.
Description check	✅ Passed	The description is directly related to the changes and accurately explains the ordered-hashtable fix and its motivation.
Docstring Coverage	✅ Passed	No functions found in the changed files to evaluate docstring coverage. Skipping docstring coverage check.
Linked Issues check	✅ Passed	Check skipped because no linked issues were found for this pull request.
Out of Scope Changes check	✅ Passed	Check skipped because no linked issues were found for this pull request.

---

Thanks for using CodeRabbit! It's free for OSS, and your support helps us grow. If you like it, consider giving us a shout-out.

❤️ Share

• X
• Mastodon
• Reddit
• LinkedIn

_{Comment @coderabbitai help to get the list of available commands.}