<@UJBPFB3SN> I hope you don't mind me asking you t...

# compose-destinations

[Rafael Costa]

@Ian Lake I hope you don't mind me asking you this here.
I was testing something with Compose Navigation and it turns out, something like what I have in below picture, will make it so "profile" composable that belongs to "graphB" will never be used. (Notice that "profile" route is used in two composable calls)
Question is: couldn't compose navigation check for route uniqueness and maybe throw an exception? I mean, this seems like a bug that the developer might not notice. One of these Composables will never be used.
I will open an issue if you agree, but I wanted to check in with you first since maybe I'm missing something.
(uploading picture since the syntax highlighting helps here more than being able to copy code from it)

[Ian Lake]

Keep in mind that each destination could still have unique deep links that would actually let you get to the second one

[Ian Lake]

And I believe there's an open issue for making the route finding start looking from where you are, rather than from the root, which would mean it could be used

[Rafael Costa]

Hmm meaning if you are already on graphB then navigating to profile would get you to graphB’s profile?

[Ian Lake]

Correct

[Rafael Costa]

That’s what I was expecting when I put this together. But still if you are in neither one it would feel a bit random. Of course at that point maybe the root doesn’t even know about profile so it would only navigate to one of the nested graphs

[Ian Lake]

It isn't random at all, it is in the order they are declared

[Rafael Costa]

The deep links thing is also a good point. But I guess we could check for same route and same deep links then, right?

[Rafael Costa]

It isn't random at all, it is in the order they are declared

are you sure? Coz I specifically tried that 😅
Let me make sure, but I believe I did. Anyway, I imagine it’s not random literally, but it may feel like that if users don’t know what’s the logic behind that decision

[Ian Lake]

Deep links are lazily evaluated, so you'd be slowing down the entire system if you were to check them all at creation time

[Rafael Costa]

I see. That makes sense. Maybe in compose destinations I can do that at compile time then 😜

[Rafael Costa]

I’ll think about it. Thank you @Ian Lake it was really helpful!

[Ian Lake]

If you can't find the issue about route ordering preferring the graph you are already on, please file a new one and I can dup it if I can find the other

[Rafael Costa]

Deal, I’ll search for it.

[Rafael Costa]

Created a new one:
https://issuetracker.google.com/issues/319671253
Hopefully I made it clear enough 😅

[Ian Lake]

Yep, looks good to me. If I do find the one that was already filed, I'll mark it as a duplicate. Otherwise, we'll just use this one. Thanks for filing the issue!