Codecov Report

✅ All modified and coverable lines are covered by tests.
✅ Project coverage is 98.47%. Comparing base (6bbd7bc) to head (30781dc).

@@            Coverage Diff             @@
##           master   #18435      +/-   ##
==========================================
+ Coverage   98.46%   98.47%   +0.01%     
==========================================
  Files         176      176              
  Lines       22831    22831              
==========================================
+ Hits        22480    22483       +3     
+ Misses        351      348       -3     

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

@andrewleech please check CI here: ruff formatting, and the unix port macos seems to lockup during the tests.

Code size report:

Reference:  samd: Convert port to use new event waiting functions. [1f601e8]
Comparison: tests: Add Ctrl-C interrupt test for the repl_ test framework. [merge of dcc243d]
  mpy-cross:    +0 +0.000% 
   bare-arm:    +0 +0.000% 
minimal x86:    +0 +0.000% 
   unix x64:    +0 +0.000% standard
      stm32:    +0 +0.000% PYBV10
      esp32:    +0 +0.000% ESP32_GENERIC
     mimxrt:    +0 +0.000% TEENSY40
        rp2:    +0 +0.000% RPI_PICO_W
       samd:    +0 +0.000% ADAFRUIT_ITSYBITSY_M4_EXPRESS
  qemu rv32:    +0 +0.000% VIRT_RV32

Thanks @dpgeorge, both fixed now. Branch is rebased and split into two commits.

The macOS hang is the same tcdrain() issue we hit with pyserial PTYs (pyserial/pyserial#822). TCSAFLUSH in mp_hal_stdio_mode_raw/mp_hal_stdio_mode_orig calls tcdrain() which blocks on macOS until the PTY master has read all pending slave output. First commit changes both to TCSANOW, I don't think the drain is doing anything useful here but happy to discuss.

The test approach is also reworked. The controlling terminal setup (setsid/TIOCSCTTY/tcsetpgrp) is now gated behind a # sigint: directive rather than applied to all repl_ tests. The PTY needs to be the child's controlling terminal for \x03 to generate SIGINT, since pyexec_friendly_repl restores original terminal settings (ISIG on) before executing user code.

Also dropped repl_ctrl_c_paste_cancel.py (already covered by repl_paste.py) and reduced the sleep from 30s to 10s.

Thanks for updating.

I've rerun the unix CI on this PR 10 times now, to test reliability. Over those 10 runs I saw only one error, on the unix qemu_arm job running the tests, cmdline/repl_emacs_keys.py failed. But I don't think that's anything to do with the changes in this PR.

On other PRs I do sometimes see failures with cmdline/repl_autocomplete_underscore.py, and maybe the changes here will resolve that.

I've seen a few jobs intermittently fail still, often possibly related to the load on the ci servers at the time, eg. #19103

Though considering repl_emacs_keys.py doesn't have a timing / timeout loop that doesn't seem as likely the cause here.