[EMRahman]

Superseded by merged PR #142. The revised docs/features/MATCH_TAGS_PLAN.md is now on main, preserving this proposal and its review fixes while extending it to in-progress matches, comparison and private notes. Both review threads on this PR are resolved; there are no separate implementation changes to merge.

Summary

Adds docs/features/MATCH_TAGS_PLAN.md, a design for tagging a finished match with:

• who was played (opponent, plus partner and second opponent in doubles), with names you can pick again later
• the Competition Type: Friendly / League / Tournament / Practice

Tags are edited on iOS and shown read-only on the watch in v1. They're used in the AI Coach prompt and in shared reports. The doc also includes a sensitive-data review.

This PR is docs only. No code changes.

Key findings recorded in the plan

• The current merge rule would silently delete tags. In MatchMergePolicy.resolve, case 4 picks the record with the later endTime and uses >=. The watch's copy of a finished match has the same endTime as the phone's, so the next full-history push from the watch replaces the phone's tagged record with an untagged one. ArchiveBackupPolicy.resolveBackup has the same shape. Tags therefore merge separately, using their own updatedAt last-write-wins clock. That Core PR must land before any UI can write a tag.
• Mixed versions: an older build re-encodes a record without tags. So nil means "no information", and clearing tags is stored as an empty MatchTags with a newer timestamp.
• Naming clash: MatchType already means singles/doubles, so the new enum is CompetitionType, matching its "Competition Type" UI label.
• Forward compatibility: the competition type is stored as a raw string. A case added later then shows as unknown on older builds instead of making the whole [MatchRecord] archive fail to decode (TECHNICAL_DEBT docs: tidy micro-site hero layout #18 and docs: reword iPhone line in hero subtitle to reduce repetition #19).
• Delivery to the watch: tag edits go on a dedicated matchTagsUpdate key. Sending a full record would go through .singleMatch, which fires completeCurrentMatchIfMatching. The phone also resends tags when watch history arrives with older or missing tags, and when a match newly appears in the watch's manifest (§6.1).
• Roster comes from the archive: the list of players is built from saved matches, so there is no second store to back up, sync or tombstone.

Sensitive-data review (§8 of the plan)

These are the app's first third-party names. Decisions:

• Names are left out of the AI prompt by default (opt-in toggle).
• Names never go in filenames or spoken announcements, and are never logged with privacy: .public.
• Names in the HTML export go through the existing scriptSafe and HTML escaping, with tests using hostile names.
• "Forget player" removes a name everywhere.
• No Contacts access.
• The privacy policy wording changes in the PR that ships the feature, not this one.
• The App Store label is expected to stay "Data Not Collected".

Owner decisions (§11)

• UI label: Competition Type; the code uses the same name (CompetitionType, MatchTags.competitionType, Meta.competitionTypeLabel)
• Practice is included as a fourth option
• Player names are on by default in shared reports, with a toggle to turn them off
• iPhone-only editing in v1; the watch shows tags read-only
• The six proposed defaults are confirmed (§11.1)

Verification

Docs only, so nothing to build or test. I checked the cited code locations (merge sites, wire handling, HTML escaping, test file names) against the current main.

🤖 Generated with Claude Code

https://claude.ai/code/session_01Q4hwsXcv3uCzZ8W2pypt4r

[chatgpt-codex-connector]

💡 Codex Review

Here are some automated review suggestions for this pull request.

Reviewed commit: cb042490c4

ℹ️ About Codex in GitHub

Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you

• Open a pull request for review
• Mark a draft as ready
• Comment "@codex review".

If Codex has suggestions, it will comment; otherwise it will react with 👍.

Codex can also answer questions or update the PR. Try commenting "@codex address that feedback".