Status Update
No update yet.
Comments
so...@gmail.com
Now, instead of crashing, we ignore invalid deep links, printing a log message saying as much.
so...@gmail.com
The problem still occurs on version 2.2.0-rc04
so...@gmail.com
il...@google.com
il...@google.com
il...@google.com
ch...@gmail.com
il...@google.com
NavDestination node = i == 0 ? mGraph : graph.findNode(destinationId);
if deepLink.length == 1 node never will be null and this will cause IllegalStateException
Bug demo:
get NavigationAdvancedSample from github
on R.id.about_btn click
NavDeepLinkBuilder(it.context)
.setGraph(R.navigation.list)
.setDestination(R.id.leaderboard)
.createTaskStackBuilder()
.startActivities()
will crash because R.id.leaderboard is start destination (findInvalidDestinationDisplayNameInDeepLink will have deepLink length == 1)
NavDeepLinkBuilder(it.context)
.setGraph(R.navigation.list)
.setDestination(R.id.userProfile)
.createTaskStackBuilder()
.startActivities()
will work because findInvalidDestinationDisplayNameInDeepLink will return NOT null
pi...@gmail.com
It still crash with the latest version 2.2.1. I think the problem lies somewhere around this code
NavDestination node = i == 0 ? mGraph : graph.findNode(destinationId);
When i is 0, just return mGraph itself. When the deep link destination is a root of bottom_nav tab, the deeplink.length becomes 1. Which means this function still says there's no invalid Destination in deep link even if deepLink[0] is totally irrelevant tomGraph.
So this step passes and when findDestination() is actually triggered down the road, it won't find any destination and thus lead to unknown destination crash.
All NavControllers of bottom nav automatically call handleDeepLink() as part of onGraphCreated(). The problem arises the moment NavController whose graph is irrelevant to the root deep link destination calls handleDeepLink().
Description
Version used: 1.0.0-alpha11
I'm seeing some strange behaviour when pressing back button and transistion back to a fragment from a fragment of the same id.
My app uses some generic fragment structures that are driven by args passed in, as such there are places like:
FragmentA -> FragmentB -> FragmentB -> FragmentC
However if the back button is pressed on the second instance of FragmentB, FragmentA is replaced in the navhost while the navgraph correctly displays in the toolbar. Forward navigation from this point however is missing from the navgraph and results in an IllegalStateException with the action 'not found' in the graph, unless the back button is pressed again (pushing the navgraph destination back to FragmentA)
Is this intended behaviour? If so is there a way to declare FragmentB as 'stackable'?