Value::try_into makes use of TryInto difficult

[djmitche]

I'd like to write some conversions for my own structs from instances of Value, and std::convert::TryFrom / TryInto are good choices for this. However, the existence of a try_into method on the struct, unrelated to these traits, shadows the trait method.

Could the try_into method be renamed into try_deserialize or something of the sort?

[matthiasbeyer]

IMO, the try_from function should be reimplemented using the TryFrom trait anyways.

I think this is a good idea!

[matthiasbeyer]

Fixed in config-rs-maint as of master def3702fac13605af303b8319af15943cc7ff8e5.

[conradludgate]

Sadly I don't think it's possible to implement it using TryFrom for all T because it conflicts.

I'm not quite sure what with, but I think it's because a type could implement From<Value> and Deserialize individually downstream and that would have conflicting implementations

[conradludgate]

Should we just go with try_deserialize() for now?

[djmitche]

that sounds good to me :)

I'm confused, though, because it looks like @matthiasbeyer already did this?

[conradludgate]

I'm confused, though, because it looks like @matthiasbeyer already did this?

Looking through their old maint fork, they only had this
https://git.sr.ht/~matthiasbeyer/config-rs-maint/tree/master/item/src/value.rs#L154. No TryFrom that I can find

[matthiasbeyer]

Yeah I even tried to implement the TryInto trait, but it was not possible when I tried... But having a function named try_into() is counter-intuitive, because it suggestes that the trait is there... which it is not. So I guess try_deserialize() it is, right?

[conradludgate]

I would say that's a consensus reached

[danieleades]

looks like this is already merged. can this be closed?

[matthiasbeyer]

Yes! Thanks for pinging me!