Unable to clone model fields in view! macro

[hwittenborn]

I'm trying to do something similar to this inside a view! macro in my codebase:

connect_changed[model.errors as errors, sender] => move |name_input| { ...

This is currently failing to compile though:

error: expected one of `:`, `;`, `=`, `@`, or `|`, found `.`
   --> src/login.rs:321:46
    |
321 |                         connect_changed[model.errors as errors, sender] => move |name_input| {
    |                                              ^ expected one of `:`, `;`, `=`, `@`, or `|`

I'm assuming this should be allowed since glib::clone! allows for it (see https://gtk-rs.org/gtk-rs-core/stable/latest/docs/glib/macro.clone.html#errors). I'm not sure how easy this would be to add support for in the parser though, but it would make my code significantly cleaner if I was able to.

[hwittenborn]

It looks like this is already handled as mentioned in #162 (comment), but it doesn't look like such behavior is currently documented.

Does the documentation for that not exist, or am I just looking in the wrong place?

[hwittenborn]

I looked here in the documentation as well, and it also doesn't appear such behavior is documented.

[hwittenborn]

Something I found a bit surprising is that using the new_var = old_var.field syntax doesn't automatically clone the field:

// Automatically clones `sender`
connect_changed[sender] => move |name_input| { ...

// Doesn't clone `model.errors`
connect_changed[errors = model.errors, sender] => move |name_input| { ...

// Clone's `model.errors` through manual `.clone()` call
connect_changed[errors = model.errors.clone(), sender] => move |name_input| { ...

Since the [var_name] syntax automatically clones the variable, I was thinking that the [new_var = old_var.field] syntax would also.

I can't really think of a case where you'd just rename with the [...] block but not clone, because at that point you could just let the closure take ownership of the old_var in question. Is this intended behavior for some other reason?

[AaronErhardt]

Something I found a bit surprising is that using the new_var = old_var.field syntax doesn't automatically clone the field:

This is because you might not want to use .clone() to duplicate your data. It's surely the most common trait, but some types in Rust have use methods.

[hwittenborn]

I was mostly wondering when you wouldn't want to though, as I was thinking you'd pretty much always want to clone when using stuff like [errors = model.errors]. I guess if you had something like model.errors.error1, then it might make sense to just pass in the entire model.errors struct into the closure, while leaving the rest of model outside of it.

That makes enough sense to me, but the currently behavior for the [var = var.field] stuff still needs documented, right?

[AaronErhardt]

Yes, documentation is missing and we could add a shorter syntax for cloning struct fields, so we can keep this issue open until this is resolved.