Skip to content

v2.0.4

Choose a tag to compare

@github-actions github-actions released this 02 Aug 22:44
· 115 commits to main since this release

Commits

  • d9f8a9f Release 2.0.4
  • 6d89754 2.0.4: the scrubber stays linear on hostile separator runs
  • 5967dda 2.0.4: close the scrubber's residual leak paths
  • 1fb45e9 2.0.4: scrub user-identifying data out of Sentry error events
  • c1274a2 F-808 step 7d: quality-review fixes — the glossary goes back to name-only
  • ad5935e F-808 step 7c: precision leftovers
  • 4a36ef5 F-808 step 7b: spec-review accuracy fixes
  • 546731a F-808 step 7: the docs describe a backend keyed by where a window can be seen
  • 002b348 Merge F-808 test pre-work: capability regate, EnumWindows helper, integration twin
  • 977566c F-808 step 10f: name the normalizer test class after the function it tests
  • 3aea184 F-808 step 10e: stage-2 review fixes — contain the ownership check's raise
  • 08002f3 F-808 step 8-pre: correct the double-exec comment — the second exec is unconditional
  • 7d08c5d F-808 step 8-pre: review fixups — the kill net must survive a failing close
  • 9c539d3 F-808 step 8-pre: ask the machine whether a headed spawn is really visible
  • 22529ad F-808 step 10d: size the record lock for the critical section it now guards
  • f4e58f3 F-808 step 10c: keep kill-orphans --force meaning what it says
  • 3fe6b37 F-808 step 8-pre: gate F-804's headed nodes on display capability, not platform
  • 38aa897 F-808 step 10b: a backend may only reap browsers nobody living owns
  • f437cab F-808 step 10a: give browser_pids.json one home, and make every write merge
  • d84323e F-808 step 6c: quality-review fixes — falsifiable remedy asserts, hermetic doctor tests
  • 69e48ad F-808 step 6b: a recorded desktop backend only silences the remedy while it is alive
  • 60e48da F-808 step 6: say that own_or_first_port's own-context branch no longer decides anything
  • 172e014 F-808 step 6: doctor reports a backend per display context
  • 58e7725 F-808 step 5c: the doomed_spawn fixture is all tripwires, as documented
  • 29f02a0 Test runs no longer ship injected failures to the real Sentry
  • d209b46 F-808 step 5b: pin the context token; state the cap raise honestly
  • 2b22fe1 F-808 step 5: a headed spawn with no displayable desktop raises instead of returning a ghost
  • 4e2ede3 F-808 step 4d: restart terminates only the port it is about to bind
  • f02334c F-808 step 4c: a proven-capable client adopts only its own desktop
  • 09433a0 F-808 step 4b: an unverified entry is never a port conflict
  • 85f7fe6 F-808 step 4: discovery prefers a window-capable backend
  • 62d4813 F-808 step 3c: quality fixes for the registry
  • 32b3185 F-808 step 3b: a v2 write supersedes the v1 entry on its port
  • 7989dee F-808 step 3: server.json records one backend per display context, v1 records still readable
  • 663da1a F-808 step 2b: quality fixes for backend_registry
  • efed9d0 F-808 step 2: move the backend record out of singleton.py
  • 8a78561 F-808 step 1b: quality fixes for display_context
  • a1b3075 F-808 step 1: display_context, the one home for "can a window be seen here"
  • 215b66c plan_F808: headed browsing works wherever a desktop exists
  • 0fbb7bd F-808: a headed spawn is invisible when the backend was cold-started from a non-interactive session
  • 4673e62 Record the Windows startup-herd hang evidence (feeds F-509)