[rtimms]

Fixes #5884.

Cause

The Example notebooks job intermittently times out because of a bug in ipykernel 7.3.0, not in PyBaMM. ipykernel 7.2 and 7.3 send shell replies with a raw send on the ROUTER socket that a ZMQStream reads requests from. That can leave an execute_request unread: the kernel sits idle at 0% CPU, never replies, and nbmake waits out its 300 s timeout (ipython/ipykernel#1554). ipykernel 7.4.0 fixes it by sending the reply through the stream (ipython/ipykernel#1529). ipykernel 7.3.0 came in with #5749 on 16 Sep.

Evidence

• The 20 A re-solve in lead-acid.ipynb takes under 1 s and is deterministic; 1,200 looped solves never stalled.

• Running 24 copies of the notebook through nbmake with 8 workers reproduced the hang locally (3/24, then 1/24). Hung kernels were idle in their event loop, and a faulthandler dump in the cell never fired, so the cell had already finished. One timeout landed on the trivial pybamm.print_citations() cell.

• 48 copies per batch:

  ipykernel / jupyter-client	Result
  7.3.0 / 8.10.0 (current lock)	1–3 hangs per 24
  7.2.0 / 8.8.0	48/48 passed
  7.4.0 / 8.10.0	48/48 passed in four batches

  One further 7.4.0 batch had 16 failures whose log was lost to a session restart, so its cause is unknown; the four clean batches came after it.

Change

• dev extra: ipykernel>=7.4.0; python_version >= '3.11'. ipykernel 7.4 requires Python 3.11+, so 3.10 stays on 7.3.0; the examples job runs on 3.13.

• uv.lock: only ipykernel changes (7.3.0 for 3.10, 7.4.0 for ≥3.11).

• The save_custom_model and load_custom_model doctests both wrote basicdfn_model.json in the working directory, which made the parallel Doctests job fail on this PR. The save_custom_model example now writes basicdfn_saved.json.

No CHANGELOG entry, as this only changes a dev dependency.

🤖 Generated with Claude Code

[github-actions]

Bencher Report

Project	PyBaMM
Branch	claude/pybamm-issue-5884-df3983
Testbed	bare-metal

⚠️ WARNING: Truncated view!

The full continuous benchmarking report exceeds the maximum length allowed on this platform.

🐰 View full continuous benchmarking report in Bencher

[codecov]

Codecov Report

✅ All modified and coverable lines are covered by tests.
✅ Project coverage is 98.13%. Comparing base (54318ca) to head (78e8e91).

Additional details and impacted files

@@           Coverage Diff           @@
##             main    #5905   +/-   ##
=======================================
  Coverage   98.12%   98.13%           
=======================================
  Files         348      348           
  Lines       35578    35578           
=======================================
+ Hits        34912    34913    +1     
+ Misses        666      665    -1     

☔ View full report in Codecov by Harness.
📢 Have feedback on the report? Share it here.

🚀 New features to boost your workflow:

• ❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.