Keystar v0.7.0

Changed

• Member audit handles large corporations: search by character, account (main) or character ID as you type, click a
  stat tile (In-game roster, Registered, Not registered, Missing or revoked ESI) to show only those characters, and page
  through 50 at a time. Searching, filtering and paging happen in the database, and the state is in the URL, so a
  filtered view can be bookmarked or shared.
• The unregistered-members count on Mining Overview and the token warnings on Users & Roles now open the member audit
  already filtered to the characters concerned.

Fixed

• Sign-in could redirect to another site when returnTo contained a tab, line feed or carriage return (for example
  /%09/evil.example). The return path is now resolved like the browser would and rejected unless it stays on Keystar.
  (#14)
• The mining CSV export no longer turns negative security status (-0.45) into text, so null-sec and wormhole rows
  stay numeric in spreadsheets. Names starting with =, +, - or @ are still neutralised.
  (#18)
• Two admins demoting or disabling each other at the same moment could both succeed and leave Keystar without an
  admin, after which the next pilot to sign in became admin. Role and access changes now re-check both users at the
  moment of the change, one at a time, and only the very first account ever is made admin automatically.
  (#16)
• When EVE's ESI is down or rate limiting, the appraisal no longer saves real items as unrecognised lines or with
  missing or stale prices: it shows "ESI unavailable, try again" and saves nothing. The ore field estimator says
  which ores couldn't be priced and tries them again, and both log the ESI error so admins can tell an outage from bad
  input. (#17)
• Name lookups no longer multiply ESI requests during an outage. Keystar splits a batch only when ESI rejects it for
  an invalid id; a server error, timeout or rate limit now fails the job, so the scheduler backs off instead of
  sending about two failing requests per id and reporting success. (#19)
• The hourly market price job no longer throws away a whole run, including the ore values the mining dashboard uses,
  when a single item can't be priced. Items that fail keep their previous values and are counted in the job summary.
  The job now prices only ores in the mining ledgers and items appraised or valued in the field estimator in the last
  14 days, rather than every item ever appraised. When a background job fails part-way, its remaining ESI requests
  stop instead of running on into the retry. (#15)
• The ESI client no longer pauses a whole rate-limit group for 15 seconds after a response that names the group but
  doesn't report its remaining tokens.

---

Docker image: ghcr.io/theragus/keystar:0.7.0

# in .env: KEYSTAR_VERSION=0.7.0
docker compose pull && docker compose up -d