Skip to content

[temp][do not merge] probe: #889 Windows status protocol (O1/O2 of the #885 closeout plan) - #890

Closed
Sunrisepeak wants to merge 1 commit into
chore/2026.11.5.1-release-closeoutfrom
probe/885-open-questions
Closed

Sunrisepeak wants to merge 1 commit into
chore/2026.11.5.1-release-closeoutfrom
probe/885-open-questions

Conversation

@Sunrisepeak

@Sunrisepeak Sunrisepeak commented Nov 6, 2026 •

Copy link
Copy Markdown
Member

Temporary probe for the open questions O1/O2 of .agents/docs/2026-11-07-ecosystem-885-closeout-design.md (tracks #885, measures #889). Do not merge.

  • Commit 1 (baseline, O2): the source of 2026.11.7.1: compiler recovery, build generations, platform version groups and minimal header regions #889 head 614227f1 unchanged; only the windows-host x86_64-linux-musl cross build of mcpp runs, repeated 20 times on one built binary.
  • Commit 2 (diagnostics, O1 and the B-class attribution): the same repetition with temporary instrumentation. On Windows each record publication also asks OpenKal about the exact temporary (located preopen, relative-name shape, read, parent stamp, root identity); every managed supervisor exit leaves a tagged line; a waiter that concludes the compiler controller is gone prints what it observed and again three seconds later.

Every other workflow is removed on this branch. Output goes to the step summary and the uploaded artifacts.

Sunrisepeak added a commit that referenced this pull request Nov 6, 2026
… process is a lost controller

On Windows the name a rename installs can be refused to a reader for a
short interval. The waiter read that refusal as an unreadable controller
record and declared a running compiler lost: 10 of 40 Windows-host cross
builds failed this way (#890). Records and statuses are now opened with a
bounded backoff (kRecordSettle, 250 ms) while the name exists; an absent
name is absent at once and a refusal that outlasts the bound is an error.
The controller's state is three-valued: only an identity that names an
ended process is ended; an unreadable record is unknown and the waiter
keeps waiting within its existing bounds. A loss is reported with the
observation that decided it. copy_atomic publishes through the host
namespace like every other record of the module, so a name in Windows'
long spelling stays valid. A cancelled managed supervisor writes its
status, and a registration failure reaches the supervisor log.

Co-authored-by: speak-agent <248744407+speak-agent@users.noreply.github.com>
… candidate (D1 acceptance)

Co-authored-by: speak-agent <248744407+speak-agent@users.noreply.github.com>
@Sunrisepeak
Sunrisepeak force-pushed the probe/885-open-questions branch from 584c985 to 2821afc Compare November 6, 2026 20:42
@Sunrisepeak

Copy link
Copy Markdown
Member Author

Probe complete: the Windows-hosted linux-musl cross build of the C1 candidate (dd53530) passed 20 of 20 repetitions (run 42290667854); before C1 the B-class failure rate was 10/40. Recorded in the #885 closeout design, §11.3.

@Sunrisepeak Sunrisepeak closed this Nov 6, 2026
@Sunrisepeak
Sunrisepeak deleted the probe/885-open-questions branch November 6, 2026 23:01
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

1 participant