Modify interface to take parameters as reference about parsec-client-rust

[ionut-arm]

@hug-dev would this impact the SE driver in a major way?

[hug-dev]

this could/should be changed so that most parameters are instead passed by reference.

what do you think would be the advantages of this? Do you have any examples in mind?

@hug-dev would this impact the SE driver in a major way?

I don't think so! It can adapt 😃

[ionut-arm]

what do you think would be the advantages of this? Do you have any examples in mind?

Yeah, the key name is a big example - we take it as an owned string everywhere... it would be much nicer and idiomatic for a user to just do &key_name instead of key_name.clone() every time. But pretty much all other attributes (e.g. AsymmetricEncryption for sign and verify) do the same

[hug-dev]

Yeah, the key name is a big example - we take it as an owned string everywhere... it would be much nicer and idiomatic for a user to just do &key_name instead of key_name.clone() every time.

Definitely agree for the name! For the other non-Copy type, I guess we will have to review them on a case-by-case basis. Basically it only depends if the client needs to re-use the values after calling the function.

[hug-dev]

I guess the default should be to take references everywhere unless the types are Copy.
For String it should be &str and for Vec<u8> &[u8]

[ionut-arm]

I think we've debated this before - passing by reference ends up "hiding" the fact that you actually want an instance of that data and that you're going to clone it. I don't know if it's sensible to just assume that if a type implements Clone we can treat a reference the same way as an owned instance.

[hug-dev]

I think most BasicClient methods take references to buffers now. The thing that would be nice is to use &str instead of String for keys as the latter implies making .clone everywhere... Ideally the clone should be done at the end to hide the complexity before.

Renaming this issue to that. Actually closing and opening a new one.