source-academy / frontend / 30760802368

45%

push

github

Thread gutter-click breakpoint lines into Conductor runs (#4241)

* Thread gutter-click breakpoint lines into Conductor runs (py-slang#383)

The Conductor /__cse_config__ virtual file only ever carried stepLimit,
so a gutter click never reached py-slang - it has no wire path to the
evaluator at all today, unlike the legacy insertDebuggerStatements
`debugger;`-insertion trick (itself a no-op under Conductor). Resolve
the active tab's ace breakpoints to 1-indexed lines and include them as
breakpointLines, matching py-slang#384's PyCseEvaluator/PyStepperEvaluator,
which resolve each line to its closest enclosing statement and treat it
exactly like an explicit breakpoint() call.

Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01MvqZCkjN8CLSBWCoNSZbAG

* Address CodeRabbit review: scope breakpointLines to editor Runs only

Two real bugs in the /__cse_config__ breakpointLines wiring (#4241):

- It applied unconditionally, so a REPL chunk (or a test-case/block-
  extra-methods run) sharing evalCodeConductorSaga would get the active
  editor tab's breakpoints spuriously attached, even though none of
  those runs use the editor's content. It also fell back to editor tab
  0 when no tab was active, rather than sending no breakpoints at all.
  Now gated on actionType === WorkspaceActions.evalEditor.type (an
  actual Run) with a genuine active tab.

- evalEditorSaga concatenates a non-empty, non-TYPED programPrependValue
  plus a newline onto the entrypoint before this saga runs, shifting
  every line of the student's own code down by the prepend's line
  count - breakpoints (recorded against the un-prepended editor value)
  now carry that same offset.

Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01MvqZCkjN8CLSBWCoNSZbAG

---------

Co-authored-by: Claude Sonnet 5 <noreply@anthropic.com>

3047 of 7788 branches covered (39.12%)

Branch coverage included in aggregate %.

6 of 7 new or added lines in 2 files covered. (85.71%)

7010 of 14531 relevant lines covered (48.24%)

30.29 hits per line