Skip to content

docs: write the [1.5.0] changelog section (TRA-1142) - #595

Open
mikestankavich wants to merge 1 commit into
mainfrom
docs/tra-1142-changelog-for-1-5-0
Open

docs: write the [1.5.0] changelog section (TRA-1142)#595
mikestankavich wants to merge 1 commit into
mainfrom
docs/tra-1142-changelog-for-1-5-0

Conversation

@mikestankavich

Copy link
Copy Markdown
Contributor

CHANGELOG.md's [Unreleased] section was empty against 104 commits / ~20 PRs since v1.4.1. TRA-1142's acceptance requires a real [1.5.0] section before the tag is cut, so this writes it.

The recount changed the release

I built this from git log v1.4.1..main, not from the v1.5.0 label — TRA-1142 explicitly asks for that ("recount from the platform repo before cutting the tag"). The label understated the release in two ways:

The schema moves to 40, not 39. 000040_users_must_change_password landed via TRA-1135 (#590), which carried no v1.5.0 label and so was invisible to the label-derived shipping list. TRA-1142's own warning — "'no migration expected' is exactly when a stray one goes unnoticed" — is precisely what happened. I've added the label to TRA-1135 and corrected the release ticket's acceptance criterion.

Four other user-visible landings were unlabelled: tag-store batching toward TRA-1150 (#582), BLE write-failure visibility (#583), fragment-reassembly hardening (TRA-1148, #581), and removal of a dead device-name search (#585).

Two deliberate wording choices

  • TRA-1150 is not claimed as fixed. perf: remove O(n²) in the tag hot path — 6.5× faster per read at 400 tags (TRA-1150) #582 is described as what it is — a main-thread starvation fix — with an explicit note that the underlying wedge is still open. TRA-1150 is still Backlog; calling it fixed in a customer-facing changelog would be wrong.
  • TRA-1120 is framed as completing 1.4.0's TRA-1108 work rather than duplicating it. 1.4.0 fixed the deep link and the mask width; this fixes the registry values that were already stored zero-stripped.

Note

Dated 2026-08-27. Bump the date if the tag slips — it's a one-word edit and it's the only thing in here that can go stale.

Refs: TRA-1142

`[Unreleased]` was empty against 104 commits since v1.4.1, and TRA-1142's
acceptance requires a real `[1.5.0]` section before the tag is cut.

Recounted from `git log v1.4.1..main` rather than from the `v1.5.0` label,
which turned out to understate the release in two ways:

- **The schema moves to 40, not 39.** `000040_users_must_change_password`
  landed via TRA-1135 (#590), which carried no `v1.5.0` label and was
  therefore invisible to the label-derived shipping list. TRA-1142 predicted
  "no migration expected is exactly when a stray one goes unnoticed" and that
  is what happened. The label has since been added to TRA-1135.
- Four other user-visible landings were unlabelled: tag-store batching toward
  TRA-1150 (#582), BLE write-failure visibility (#583), fragment-reassembly
  hardening (TRA-1148, #581), and removal of a dead device-name search (#585).

TRA-1150 is described as what it is — a main-thread starvation fix that does
not close the ticket, which is still Backlog. Claiming the wedge fixed here
would be wrong.

Dated 2026-08-27; bump it if the tag slips.

Refs: TRA-1142
@github-actions

Copy link
Copy Markdown
Contributor

🚀 Preview Deployment Update

✅ This PR has been successfully merged into the preview branch.

The preview environment will update shortly at: https://app.preview.trakrf.id

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

1 participant