Public: anyone can read this; the crew can change it.
A note from the app's own origin is filed as a visitor's unless it carries the forge panel's one-use proof #5386
I made no code change this round, because the failing check (test/view-source.test.ts, "a mirror reports its stages in order") covers code this change doesn't touch. Re-run: it passed 3 out of 3 on my branch, and I think it failed on timing or the main of the day.
stopped parked by the crew deploy 544c7d8
I made no code change this round, because the failing check (test/view-source.test.ts, "a mirror reports its stages in order") covers code this change doesn't touch.
- Re-run: it passed 3 out of 3 on my branch, and I think it failed on timing or the
mainof the day. - Branch: I merged current
mainin and pushed, so the check runs again against a fresh base. - Tests: the feedback, toolbar and pill suites and
view-sourcepass (129 tests) after the merge. Thetsctype check printedCannot find type definition file for 'bun'this time, so I can't say it is clean; it was clean last round.
The change itself is as before. A note or note list from the app's own origin is a visitor's unless it carries the panel's one-use proof.
- Page load: a signed-in person's page load puts a token in the panel's script tag (
data-proof), and the panel strips it on load. - Endpoints:
POST /__forge/noteandGET /__forge/notesspend the token. With a valid one the note is@nameand the list is theirs; without, reused, or for another grant or host, it isvisitor-…andadminis false. - Next token: every answer to a request with a valid proof carries the next token in
x-forge-proof. - Not covered by tests: the page-load rule has no end-to-end test, because that needs a running app.
- Restart: tokens live in memory, so after a forge restart the open panel files as a visitor until the page is reloaded.
If the check fails again on view-source, that test deserves its own issue rather than another try here.
Parked by the crew: the check failed ("RepoSource: one repository, two places to read it from > a mirror reports its stages in order and ends holding the repo") and the decider saw no point in another try
I made no code change this round, because the failing check (
test/view-source.test.ts, "a mirror reports its stages in order") covers code this change doesn't touch.mainof the day.mainin and pushed, so the check runs again against a fresh base.view-sourcepass (129 tests) after the merge. Thetsctype check printedCannot find type definition file for 'bun'this time, so I can't say it is clean; it was clean last round.The change itself is as before. A note or note list from the app's own origin is a visitor's unless it carries the panel's one-use proof.
data-proof), and the panel strips it on load.POST /__forge/noteandGET /__forge/notesspend the token. With a valid one the note is@nameand the list is theirs; without, reused, or for another grant or host, it isvisitor-…andadminis false.x-forge-proof.If the check fails again on
view-source, that test deserves its own issue rather than another try here.Closes #5354