fix(core): DSRunner memory_efficient output/monitors; eager check bound validation

[chaoming0625]

Fresh review of top-level glue (dynsys, runners, delay, mixin, transform, check, shims).

• Critical — DSRunner(memory_efficient=True) returned None instead of model outputs (list.append inside tree_map); now accumulates and stacks per-step outputs on a leading time axis.
• High — memory_efficient monitors came out time-major vs the standard path's batch-major in BatchingMode; now axis-matched.
• High — check.is_float/is_integer bound checks never raised eagerly (a pure_callback swallowed the raise for concrete predicates); added a concrete fast-path that raises directly, preserving the deferred path under tracing.

⚠️ Blast radius: check.is_float/is_integer with bounds now actually reject invalid input across all call sites (valid defaults verified to still construct). In-scope: 441 passed. Findings: docs/issues-found-20260619-toplevel-glue.md.

Summary by Sourcery

Fix DSRunner memory-efficient output and monitor layout while tightening eager bound-checking behavior.

Bug Fixes:

• Ensure DSRunner(memory_efficient=True) returns time-stacked model outputs instead of None, matching the standard path for both scalar and pytree outputs.
• Align memory-efficient monitor axis ordering with the standard batch-major layout in BatchingMode so monitor shapes no longer depend on the memory_efficient flag.
• Make is_float and is_integer min_bound/max_bound checks raise eagerly for out-of-bound values while preserving deferred in-jit error signaling.
• Update jit_error and jit_error_checking_no_args to raise synchronously for concrete predicates, and adjust tests to reflect the new error behavior.
• Prevent DSRunner tests from leaking a modified global dt across test cases by restoring the default dt after each test.

Documentation:

• Add an audit report documenting identified issues and their fix/record status for top-level glue components.

Tests:

• Add regression tests covering DSRunner memory_efficient outputs and monitor shapes, including batched models and update methods that return None.
• Add regression tests asserting eager bound checks and jit_error helpers raise or not raise as expected in both eager and jitted contexts.
• Strengthen existing coverage tests for check.* utilities to expect the new eager-raising semantics.

[sourcery-ai]

Reviewer's Guide

Fixes DSRunner’s memory_efficient output stacking and monitor axis layout, and restores eager bound-checking semantics in check utilities while adding regression tests and documenting audit findings.

File-Level Changes

Change	Details	Files
Fix memory_efficient DSRunner output accumulation to return time-stacked model outputs and keep None when update() returns no value.	Replaced per-step tree_map list.append accumulation with a flat Python list of outputs and a final tree_map+jnp.stack over time Ensured empty index ranges return (None, None) instead of malformed pytrees Standardized stacking to a leading time axis using bm.as_jax/jnp.stack to match the non-memory-efficient path	brainpy/runners.py
Align memory_efficient monitor layout with standard batch-major monitors for batched models.	During memory_efficient predict() finalization, detect BatchingMode with data_first_axis='B' and move the time axis behind batch for collected monitors Convert monitor lists to NumPy arrays and use np.moveaxis for ndim>=2 to yield consistent (B, T, ...) layouts with the standard path	brainpy/runners.py
Add eager fast-paths for jit_error and jit_error_checking_no_args so bounds checks actually raise for invalid values while preserving deferred behavior under JIT.	In jit_error, short-circuit concrete non-tracer predicates by evaluating pred eagerly and calling the error function directly when True In jit_error_checking_no_args, short-circuit concrete predicates by raising the provided Exception immediately when True and returning otherwise Adjusted the pure_callback true branch helper to accept *args to be compatible with JAX callbacks	brainpy/check.py
Update and extend tests to cover DSRunner memory_efficient behavior, monitor axes, and eager bound-check semantics, and to avoid global dt leakage between tests.	Introduce a _DtRestoreMixin in dyn_runner_test to snapshot and restore the global dt after each test, and mix it into TestDSRunner/TestMemoryEfficient Add TestMemoryEfficient cases covering scalar and pytree outputs, monitor equality, None-returning update(), and batched monitor axis layout Add TestBoundChecks to verify is_float/is_integer min/max bounds raise eagerly, within-bounds values pass, and jit_error_checking_no_args behaves correctly both eagerly and under jax.jit Update existing check_coverage_test tests to expect exceptions for out-of-bound values and concrete True predicates now that errors propagate eagerly	brainpy/dyn_runner_test.py brainpy/check_test.py brainpy/check_coverage_test.py
Record audit findings for the top-level glue audit, including this PR’s fixes and remaining low/medium issues, in documentation.	Add docs/issues-found-20260619-toplevel-glue.md summarizing critical, high, medium, and low findings, marking which are fixed vs recorded-only Document specific IDs (P14-C1, P14-H1, P14-H2, P14-M2, P14-L1-4) with rationale, repro snippets, and status for future reference	docs/issues-found-20260619-toplevel-glue.md

---

Tips and commands

Interacting with Sourcery

• Trigger a new review: Comment @sourcery-ai review on the pull request.
• Continue discussions: Reply directly to Sourcery's review comments.
• Generate a GitHub issue from a review comment: Ask Sourcery to create an
  issue from a review comment by replying to it. You can also reply to a
  review comment with @sourcery-ai issue to create an issue from it.
• Generate a pull request title: Write @sourcery-ai anywhere in the pull
  request title to generate a title at any time. You can also comment
  @sourcery-ai title on the pull request to (re-)generate the title at any time.
• Generate a pull request summary: Write @sourcery-ai summary anywhere in
  the pull request body to generate a PR summary at any time exactly where you
  want it. You can also comment @sourcery-ai summary on the pull request to
  (re-)generate the summary at any time.
• Generate reviewer's guide: Comment @sourcery-ai guide on the pull
  request to (re-)generate the reviewer's guide at any time.
• Resolve all Sourcery comments: Comment @sourcery-ai resolve on the
  pull request to resolve all Sourcery comments. Useful if you've already
  addressed all the comments and don't want to see them anymore.
• Dismiss all Sourcery reviews: Comment @sourcery-ai dismiss on the pull
  request to dismiss all existing Sourcery reviews. Especially useful if you
  want to start fresh with a new review - don't forget to comment
  @sourcery-ai review to trigger a new review!

Customizing Your Experience

Access your dashboard to:

• Enable or disable review features such as the Sourcery-generated pull request
  summary, the reviewer's guide, and others.
• Change the review language.
• Add, remove or edit custom review instructions.
• Adjust other review settings.

Getting Help

• Contact our support team for questions or feedback.
• Visit our documentation for detailed guides and information.
• Keep in touch with the Sourcery team by following us on X/Twitter, LinkedIn or GitHub.

[sourcery-ai]

Hey - I've found 3 issues

Prompt for AI Agents

Please address the comments from this code review:

## Individual Comments

### Comment 1
<location path="docs/issues-found-20260619-toplevel-glue.md" line_range="9-10" />
<code_context>
+
+Branch: fix/audit-20260619-toplevel-glue
+
+Counts: Critical 1, High 2, Medium 1, Low 4. Fixed 4 (C1, H1, H2 + the M2 test-fragility);
+recorded-only 4 (M2 source-level, L1, L2, L3, L4).
+
+| ID | Severity | File | Status |
</code_context>
<issue_to_address>
**issue (typo):** Recorded-only count does not match the listed items.

The text says "recorded-only 4" but lists five items: "M2 source-level, L1, L2, L3, L4." Please update either the count or the list so they match.

```suggestion
Counts: Critical 1, High 2, Medium 1, Low 4. Fixed 4 (C1, H1, H2 + the M2 test-fragility);
recorded-only 5 (M2 source-level, L1, L2, L3, L4).
```
</issue_to_address>

### Comment 2
<location path="docs/issues-found-20260619-toplevel-glue.md" line_range="117-119" />
<code_context>
+  *concrete* (eager) predicate, `jax.pure_callback` only raises when the staged computation is
+  executed — under an eager `cond` the callback's exception is never surfaced synchronously, so
+  the function returns normally. (`true_err_fun(arg, transforms)` also had the wrong arity for a
+  no-operand callback.) Result: every eager out-of-bound check is a no-op.
+- Why it's a bug: parameter validation across the codebase silently accepts invalid values, e.g.
+  `bp.dnn.Dropout(prob=2.0)` / `prob=-1`, negative frequencies, non-positive integer sizes,
</code_context>
<issue_to_address>
**nitpick (typo):** Consider changing 'out-of-bound' to 'out-of-bounds' for standard phrasing.

In the final sentence of this bullet, update the phrase so it uses the standard 'out-of-bounds' wording.

```suggestion
  executed — under an eager `cond` the callback's exception is never surfaced synchronously, so
  the function returns normally. (`true_err_fun(arg, transforms)` also had the wrong arity for a
  no-operand callback.) Result: every eager out-of-bounds check is a no-op.
```
</issue_to_address>

### Comment 3
<location path="docs/issues-found-20260619-toplevel-glue.md" line_range="239" />
<code_context>
+  `math/object_transform/variables.py` (out of scope); **already fixed** there
+  (`size_without_batch` now returns `(10,)` for a `(4,10)` batched Variable). Verified.
+
+left_unfixed_chm: none in scope. H-45's fix lives outside this scope (variables.py) and is
+already applied upstream.
</code_context>
<issue_to_address>
**suggestion (typo):** Possible typo in 'left_unfixed_chm'.

The name 'left_unfixed_chm' is unclear and looks like a typo or unclear abbreviation. Please rename it to something more descriptive and correctly spelled (e.g., `left_unfixed_issues`).
</issue_to_address>

---

Sourcery is free for open source - if you like our reviews please consider sharing them ✨

• X
• Mastodon
• LinkedIn
• Facebook

_{Help me be more useful! Please click 👍 or 👎 on each comment and I'll use the feedback to improve your reviews.}

[sourcery-ai]

issue (typo): Recorded-only count does not match the listed items.

The text says "recorded-only 4" but lists five items: "M2 source-level, L1, L2, L3, L4." Please update either the count or the list so they match.

Suggested change

	Counts: Critical 1, High 2, Medium 1, Low 4. Fixed 4 (C1, H1, H2 + the M2 test-fragility);
	recorded-only 4 (M2 source-level, L1, L2, L3, L4).
	Counts: Critical 1, High 2, Medium 1, Low 4. Fixed 4 (C1, H1, H2 + the M2 test-fragility);
	recorded-only 5 (M2 source-level, L1, L2, L3, L4).

[sourcery-ai]

nitpick (typo): Consider changing 'out-of-bound' to 'out-of-bounds' for standard phrasing.

In the final sentence of this bullet, update the phrase so it uses the standard 'out-of-bounds' wording.

Suggested change

	executed — under an eager `cond` the callback's exception is never surfaced synchronously, so
	the function returns normally. (`true_err_fun(arg, transforms)` also had the wrong arity for a
	no-operand callback.) Result: every eager out-of-bound check is a no-op.
	executed — under an eager `cond` the callback's exception is never surfaced synchronously, so
	the function returns normally. (`true_err_fun(arg, transforms)` also had the wrong arity for a
	no-operand callback.) Result: every eager out-of-bounds check is a no-op.

[sourcery-ai]

suggestion (typo): Possible typo in 'left_unfixed_chm'.

The name 'left_unfixed_chm' is unclear and looks like a typo or unclear abbreviation. Please rename it to something more descriptive and correctly spelled (e.g., left_unfixed_issues).