discrete_dp review overlay recommends qe.tic/qe.toc, contradicting qe-code-005 and pointing at a deprecated API

[mmcky]

One reviewer action in the discrete_dp overlay tells an author to adopt qe.tic/qe.toc. That is wrong twice over: it contradicts the rule it cites, and it points at the API that QuantEcon/QuantEcon.py#833 is deprecating.

The finding

reviews/lecture-python-advanced.myst/discrete_dp.json, actions[6]:

Sweep the remaining single-instance items: … swap the three %timeit calls at 844-846 for qe.tic/qe.toc (qe-code-005), …

That surfaces in the generated report at lectures/lecture-python-advanced.myst/discrete_dp.md:67.

It contradicts the rule it cites. lectures/spec.md:64 defines qe-code-005 as "Use quantecon.timeit for benchmarking", and tools/qestyle_rules.py:1310 emits the hit text "hand-rolled benchmark loop — use qe.timeit". The correct recommendation for three %timeit calls is qe.timeit, not the Matlab-like timers.

It points at a deprecated API. QuantEcon/QuantEcon.py#833 makes tic, tac, toc and loop_timer emit a DeprecationWarning — deprecated at 0.12.0, removal targeted at v1.0 — as step 1 of QuantEcon/QuantEcon.py#786. An author who followed this action today would write code that warns on every call, and would have to redo the work at step 2.

Suggested fix

Edit actions[6] in reviews/lecture-python-advanced.myst/discrete_dp.json to name qe.timeit, then regenerate the report with tools/qestyle_draft.py per UPDATE.md — the .md is generated, so it should not be hand-edited.

One thing to confirm rather than assume: the overlay carries "source": {"commit": "b83d6da…", "blob": "845cbae…"}. Correcting the prose does not change which blob was judged, so my reading of the provenance convention is that the stamp stays as it is and this is not a re-review. Worth a maintainer's confirmation before the edit lands, since qestyle_status.py joins on it to tell fresh overlays from stale ones.

Scope

This is the only instance. I swept the whole repo for qe.tic/qe.toc/tic/toc mentions; besides this overlay and its generated report, the only other hit is lectures/spec.md:63, which reads "Mechanical (detect time.time() / tic/toc)" — correct, since that describes what the detector looks for rather than what an author should write. tools/qestyle_rules.py:1264 and tools/qestyle_report.py:96 both prescribe qe.Timer correctly.

Context

Found while auditing whether QuantEcon/QuantEcon.py#833 would affect the lecture series. It does not — all nine repos in the lectures workspace manifest are clean of the deprecated timers, verified with three independent grep engines. This overlay and one note in QuantEcon/QuantEcon.manual (tracked at QuantEcon/QuantEcon.manual#151) were the only two places left in the org still steering authors to the old API. The canonical QuantEcon/style-guide rule qe-code-004 is already correct, listing qe.Timer() as ✅ and tic()/toc() as ❌.

🤖 Generated with Claude Code