[ASDAlexander77]

More -mm=own gaps from compiling native_ts_graphics (TSLANG 71-81, written up in TSLANG_ISSUES.md).

Fixed

• 71: a value narrowed from T | null and kept by a constructor can be used after it.
• 72: a local kept by a constructor or pushed, then read. This includes a local declared in the loop, or filled in place before the move.
• 73: a private field that only the constructor writes is not overwritten by a later call, callbacks included.
• 74: cond ? null : f(), with f from another module, pushed where not null, then used.
• 75: a call through a global no store writes, initialized with a function (const f = Math.fround), is a call of that function.
• 76: a constructor that gives this to a call (this.reset()) still keeps a parameter, in a private field only it writes.
• 77: an array kept by a constructor, then filled through the local, fills the object's array.
• 78: a value pushed on, or stored into, a field of a local object made by new, then used, reads that field.
• 79: let x = ...; if (x) this.xs.push(x); return x; reads the element where the paths join, or null.
• 81: a value moved on some paths only is released on the edges that did not move it. A read after the paths join is still refused, a copyable value's read included.

Not fixed

• 80 is open. A first fix copied an owned parameter at a field store; it leaked (3M calls grew to 2.7 GB) and is backed out in this branch. own_err_string_kept_param_known says why.

Tests

• Every model: 00tslang_issue_71, 00tslang_issue_73, and 00tslang_issue_75 to 00tslang_issue_79, plus 00tslang_issue_81.
• New own error tests:
  • own_err_kept_local_array_popped
  • own_err_private_field_written_callback
  • own_err_const_alias_drops
  • own_err_kept_field_rebound
  • own_err_kept_array_filled_after_pop
  • own_err_local_object_field_popped
  • own_err_pushed_let_popped
  • own_err_moved_on_some_paths_used
• TSLANG 74's case is in own-facts-objects.cmake.
• Existing own error tests:
  • Five are still refused, at the same position, as a field borrow, and their expected text says so.
  • Three are now accepted (a value moved on some ways, an aliased store); they became cases in the 78 and 81 tests.
  • own_err_declared_drops now lends a public field, because the private one is a field 73 keeps.

Validation

• ctest: 7400 of 7400 passed.
• DefaultLib tests.ps1: all four models passed. Results are in a comment.
• 73 and 81 change releases, so I ran leak probes too: memory stays flat over 20M+ iterations.
• ntg sweep: 615 errors in 115 failing sources before this branch, 129 in 39 after. The ntg-side changes are on ntg branch own-sweep-fixes.

🤖 Generated with Claude Code

[ASDAlexander77]

DefaultLib tests.ps1 (Windows, release + debug, compile + jit), built with the branch head's tree (565f8d8b):

== gc
Finished release, compile : 224/225 passed, 1 skipped
Finished release, jit : 224/225 passed, 1 skipped
Finished debug, compile : 224/225 passed, 1 skipped
Finished debug, jit : 224/225 passed, 1 skipped
Done. All tests passed.
== none
Finished release, compile : 221/225 passed, 4 skipped
Finished release, jit : 221/225 passed, 4 skipped
Finished debug, compile : 221/225 passed, 4 skipped
Finished debug, jit : 221/225 passed, 4 skipped
Done. All tests passed.
== rc
Finished release, compile : 221/225 passed, 4 skipped
Finished release, jit : 221/225 passed, 4 skipped
Finished debug, compile : 221/225 passed, 4 skipped
Finished debug, jit : 221/225 passed, 4 skipped
Done. All tests passed.
== own
Finished release, compile : 219/225 passed, 6 skipped
Finished release, jit : 219/225 passed, 6 skipped
Finished debug, compile : 219/225 passed, 6 skipped
Finished debug, jit : 219/225 passed, 6 skipped
Done. All tests passed.

ctest: 100% tests passed, 0 tests failed out of 7400