Anyone had an issue with `stringResource` when using the formatArgs? For example `stringResource(Res...

# compose

[Stan Kocken]

Anyone had an issue with stringResource when using the formatArgs?
For example stringResource(Res.string.mystring, param1, param2) .
Sometimes it will return a previous instance of it, not with the new value of param1. Tested on the latest v1.6.0-rc01 but still present.
Looking at the source code I don't even understand how it works:

fun stringResource(resource: StringResource, vararg formatArgs: Any): String {
    val resourceReader = LocalResourceReader.current
    val args = formatArgs.map { it.toString() }
    val str by rememberResourceState(resource, { "" }) { env ->
        loadString(resource, args, resourceReader, env)
    }
    return str
}

the "args" are only used within the remember, and this line is not called when I have the bug in my case.

[Joel Denke]

Looks like a bug to me. I think should be:
val args = remember(formatArgs) {
formatArgs.map ...
}
However I am not entirely sure. Is your params dynamic and shifts value, like from an uiState or such?

[Stan Kocken]

My params are regular String, coming from a list of items…
But I really don't get why sometimes it works, sometimes it does not. And of course I'm not able to reproduce it within a smaller project to open an official bug on the bug tracker.

[Joel Denke]

If its a list I think because lists are unstable in compose. Depends when compose trigger new recompositions.

[louiscad]

Did you open an issue on GitHub about that?
It looks like an implementation bug to me.

[Stan Kocken]

@louiscad I did not, I was looking if someone had an idea why this was the case. But yes, I will do it now.

[Stan Kocken]

Just did https://github.com/JetBrains/compose-multiplatform/issues/4325

[Stan Kocken]

@louiscad, thanks for pushing me to open the issue, it was indeed a bug. https://github.com/JetBrains/compose-multiplatform/pull/4333

[louiscad]

Fun thing is that @zsmb and @Sebastian Aigner encountered it in their last stream as well, and found the exact same remember without the right key mistake.

[zsmb]

I spent about 15 minutes yesterday creating a clean sample and nice description for the issue too, then as I went to link the source code where the key was missing, I was surprised to see that it's no longer missing, and the file was updated 3 hours prior 😀

[louiscad]

Nice thing is they added a test for that, and that's a good example if you want to submit failing tests for any future bugs (I don't hope for those though)