stoically

> I just end up unwraping a lot. And then back tracking and discovering where the unwraps can and cannot fail. That's a strong argument to live with less DRY...

Started to look into an enum style `Node` implementation, and it actually isn't that bad in terms of duplication. Seems much cleaner.

Is your rewrite public somewhere? I'd be interested to take a look! Here's something I was testing, which is based on using `#[serde(default = "..")]`. Could certainly use some ergonomic...

Thanks for the pointer. It seems that the core approach is still to operate on string-based config keys, unless I'm missing something? Would you say that doing string-based operations is...

"string-based config keys" wasn't well phrased. What I actually meant was "string-based getters". Which isn't entirely true anymore for the rethink rewrite, since it introduces the concept of accessors, so...

Because, one big part of the config crate is the ability to parse from a multitude of sources in an easy way - combining that with `serde` for the concrete...

Thanks for taking the time to write down your thoughts around this. Could you maybe expand on how 1. would look like for the "concrete type with type safe defaults"-scenario...

I see, thanks for elaborating. But wouldn't that still mean that there's no type safety, given that I can change the fields of `MyConfig` without having to change `MyConfigDefault`? E.g....

Right, my bad. Attached the log to the initial description. The problematic line seems to be > [06:39:39] [DBG] [52] Jellyfin.Api.Controllers.DynamicHlsController: Starting transcoding because currentTranscodingIndex=null

It's exactly the same issue as described in https://github.com/jellyfin/jellyfin/issues/7295, hence why I mentioned a regression and linked the issue. Sorry for not being clear about it. Please let me know...