The Default impl for Command has unnecessary bounds

[Makefile-dot-in]

The Default impl on the Command struct has the following bounds, autogenerated by derivative:

impl<U, E> Default for Command<U, E>
where U: Default,
      E: Default

However, U and E don't actually need to have a Default implementation, since none of the fields in Command require construction of either to have a Default impl, and there isn't any reason why they ever would in the future - user data isn't stored in Command and E is only returned as the Err variant of Result, where constructing the Ok variant is always possible to get a default value. It would also be silly in most cases for the error to have a Default impl, and most error types don't, making this impl useless for nearly all cases.

My usecase for this is to have a command for which the subcommands are determined from a const variable in a different module - something like this:

// lib.rs
pub const ADMIN_COMMANDS: &[fn() -> poise::Command<Data, Error>] = &[ 
    log::log,
    remind::reminders_adm,
];

// admin.rs
use poise::Command;

pub fn admin() -> Command<Data, Error> {
    Command {
       name: "admin",
       subcommands: ADMIN_COMMANDS
           .map(|f| f())
           .collect(),
       slash_action: Some(|_| Box::pin(async move { Ok(()) })),
       subcommand_required: true,
       ..Default::default()
    }
}

whereas now I have to do something like this:

use poise::{Command, Context};

pub fn admin() -> Command<Data, Error> {
   #[poise::command(slash_command, rename = "admin")
   fn inner(_: Context<'_, Data, Error>) -> Result<(), Error> { Ok(()) }
   Command {
     subcommands: ADMIN_COMMANDS
        .map(|f| f())
        .collect(),
       subcommand_required: true,
      ..inner()
   }
}

I suppose in this case this isn't that big of a deal though (though it might matter more in other cases).

[kangalio]

Good point. I'm open for PRs