chadtmiller.com/knotless

Switch repository

Loading…

What changes

An account found to have no handle shows its DID for ten minutes rather than a day, then is looked up again behind the page showing it. A handle that is found is still kept for a day.

Why

An account can claim a handle just before its DNS goes live. The identity event looks it up at once and finds nothing, and the pages then showed the DID for a day, with no later event to correct it. demo.knotless.dev did this today. Ten minutes matches how long a lookup of a handle's host is kept.

An account whose handle stays broken is looked up at most every ten minutes while a page shows it, rather than once a day. Each lookup reads its DID document and resolves its handle, behind the page, so nothing waits on it.

How it was checked

forge.rs sets a row to none just now and at just over ten minutes ago: the first keeps none and the second finds the handle. Left to test on the runner.

🤖 Generated with Claude Code

This keeps a "no handle" row for ten minutes instead of a day, then looks it up again behind the page, as a found handle already is. The change in web.rs is correct. The row is stamped before the spawn, so concurrent pages share one lookup. If the DID document can't be read, rename leaves the row alone. The docs page is updated. The test sets a row to none just now and at 601 seconds ago, as the body says. I didn't see the test run, and I didn't read renamed or kept in forge.rs, so I'm assuming renamed waits for the looked-up handle to be kept. The change file isn't needed because only knotless-serve changes.