source-academy / py-slang / 30663180452

86%

push

github

Add conductor-level multi-chunk tests for PyCseEvaluatorBase and bump conductor to 0.8.3 (#373)

* Add conductor-level multi-chunk tests for PyCseEvaluatorBase

Py2JsEvaluator.test.ts and PyPvmlEvaluator.test.ts already drove their
evaluators through multiple evaluateChunk() calls, but there was no
equivalent for the CSE machine's PyCseEvaluatorBase - the persistence
mechanism a real REPL session (source-academy/py-slang#359) depends on,
once @sourceacademy/conductor's BasicEvaluator.startEvaluator loops on
further chunks instead of stopping after the first (source-academy/conductor#66).

Covers: variable/function persistence across chunks, an erroring chunk
reporting via sendError without ending the session, and two CSE-specific
concerns the shared evaluateChunk() code path handles per-chunk - snapshot
collection running against a control/stash reinitialized for that chunk
alone, and breakpoint() step indices not leaking from one chunk's snapshot
batch into the next.

* Fix prettier formatting (format:ci CI failure on #373)

One line exceeded print width - yarn format wraps the call across lines.

* Bump @sourceacademy/conductor to 0.8.3

Picks up source-academy/conductor#66 (BasicEvaluator.startEvaluator now
loops on further chunks instead of stopping after the first), published as
part of the REPL-persistence work for #359. Verified: full test suite
(71 passed, 2 skipped opt-in suites, unaffected) and typecheck clean
against the new version.

4509 of 5651 branches covered (79.79%)

Branch coverage included in aggregate %.

10017 of 11240 relevant lines covered (89.12%)

172425.44 hits per line