You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
{{ message }}
Repository navigation
a later link drops a pending pubky sign-in request #1545
Bitkit holds only one deferred scan. Two cases drop a Pubky sign-in link:
A sign-in link opened on a cold start waits behind the PIN screen. A bitcoin: URI that arrives before unlock replaces it, and the sign-in request is lost:
17:00:05.131 DEBUG [AppViewModel.kt] Replacing deferred scan from 'deeplink'
A payment URI that arrives while the Pubky Authorize sheet is open replaces that sheet with Send.
Seen while gating #1514 at d790f34. That PR does not change this behaviour.
Expected
A sign-in request already on screen, or waiting for unlock, is not discarded by a later link. The later link runs after the sign-in sheet closes, as it already does during the restore wait (Queuing 'deeplink' scan for deferred handling).
Steps to reproduce
Turn PIN on. Force-stop Bitkit and open a fresh pubkyauth://signin_grant link (adb shell am start -a android.intent.action.VIEW -d '<url>' to.bitkit.dev).
While the PIN screen is up, open bitcoin:<address> the same way.
Unlock. Send opens, and the sign-in request never appears.
Separately, with the Authorize sheet open, open bitcoin:<address>. The sheet is replaced by Send.
Root cause
AppViewModel.deferLockedScan / deferredScan hold a single slot that the next deferred scan overwrites. A non-mustComplete scan replaces the current sheet.
Suggested fix
Queue deferred scans in order, or keep a pending sign-in until it is handled. Do not replace an open Sheet.PubkyAuth with a payment scan; queue the scan until the sheet closes.
@jvsena42 thanks, this is the same pair of cases as #1542, filed a few minutes earlier from your #1514 device-test notes. Closing this one as a duplicate; #1542 is in 3.0.5.
Activity
ovitrif commentedon Oct 16, 2026
@jvsena42 thanks, this is the same pair of cases as #1542, filed a few minutes earlier from your #1514 device-test notes. Closing this one as a duplicate; #1542 is in 3.0.5.