Repository navigation
CI: "tests / zephyr" job is failing on the BLE branches — cause not established #11
Description
The tests / zephyr job of "Build CI" is red on the Pico 2 W BLE branch (run 34253440312), so the mpy-cross artifacts on that branch come from an otherwise-failing run.
I have not established the cause. From a glance at the log it dies around the native_sim port setup installing i386 SDL dependencies (libsdl2-dev:i386, libgpg-error0:i386 …), which looks like environment/infra rather than our changes. But that is a guess, and two of the changes on these branches are port-wide and could plausibly break a test:
ports/zephyr-cp/prj.confgainedCONFIG_NVS_INIT_BAD_MEMORY_REGION=y(affects every board)ports/zephyr-cp/cptools/compat2driver.pygained one mapping entry
First thing to check: whether tests / zephyr is also red on main / upstream. If it is, this is pre-existing and unrelated. If it is green there, it's ours and needs bisecting.
Tracking this so a red CI job on the BLE stack isn't quietly accepted.
Activity
tyeth commentedon Sep 8, 2026
Scope note: this issue is specifically the
tests / zephyrjob (native_sim, i386 SDL setup), whose cause is still not established.The
ports (zephyr-cp) / board (...)link failures on the same runs are a different problem with a known cause, now tracked separately as #16. Please don't conflate them — I did briefly, and it obscured that one is diagnosed and the other isn't.tyeth commentedon Sep 10, 2026
No longer reproducible — closing without a root cause, deliberately.
Build CInow passes fully green including thezephyr-tests / zephyrjob on branches cut from currentmain: #5 (run 34282151673, 41/41) and #17 (run 34281847143, 41/41).The original red run predates several
mainmerges, and the failure signature pointed at thenative_simi386 SDL dependency setup rather than anything in this stack, so an upstream fix is the most likely explanation. Not worth further bisecting given it passes now.The two port-wide changes this issue flagged as possible causes —
prj.confgainingCONFIG_NVS_INIT_BAD_MEMORY_REGION=yand thecompat2driver.pyentry — are both present on the branches that now pass, so neither was responsible.Reopen if it recurs.
🤖 Generated with Claude Code
Record the current CI assets, and the cause of the missing 10.x-mpy