Solidity: Replace _ with body in modifiers

[axic]

To make the code a bit less cryptic:

modifier owneronly {
    if (address != msg.sender) {
        throw;
    }

    body;
}

(Follow up of #535)

[redsquirrel]

So 'body' will be a keyword?

[obscuren]

How about "yield"? It's a common used keyword and generally well understood.

On 06 Sep 2016, at 06:02, Dave Hoover [email protected] wrote:

So 'body' will be a keyword?

—
You are receiving this because you are subscribed to this thread.
Reply to this email directly, view it on GitHub, or mute the thread.

[chriseth]

I don't think yield matches this use-case. Furthermore, I think we should stay with using a symbol. Admittedly, _ is not really "visible" (depending on your font), but words do not stand out as well, I would say.

The proper way to do this would be:

modifier owneronly[f] {
     if (address != msg.sender) {
        throw;
    }
    f();
}

and this also allows us to do

modifier owneronly[body] {
     if (address != msg.sender) {
        throw;
    }
    body();
}

and modifier owneronly would be identical to modifier owneronly[_] at which point we could even achieve backwards-compatibility.

(in all cases, the fact whether () should or has to be specified is debatable)

[axic]

@obscuren I like yield but it could mislead into thinking that coroutines are supported.

[axic]

@chriseth your proposal is nice, but how many people would actually use it? The default behaviour should reduce the possibility of errors.

[chriseth]

@axic we can report a warning if [] is not used.

[VoR0220]

I still say there's nothing wrong with the _.

[axic]

@axic we can report a warning if [] is not used.

I think it could be an acceptable way to go this way. Though we could start issuing a warning for _ and switch to body in the next breaking.

[VoR0220]

How would it look with param arguments with modifiers in this proposed schema?

[axic]

I would like to revive this discussion.

[leonardoalt]

I don't think yield matches this use-case. Furthermore, I think we should stay with using a symbol. Admittedly, _ is not really "visible" (depending on your font), but words do not stand out as well, I would say.

The proper way to do this would be:

modifier owneronly[f] {
     if (address != msg.sender) {
        throw;
    }
    f();
}

Instead of making the modifier explicitly mention the function, how about the other way around? That is, a modifier states a condition, and the function decides how to use it, for example as a pre or post condition.

modifier owneronly {
  require(owner == msg.sender);
}
modifier magic(uint x) {
    require(x == 0xcafe)
}
function f(uint x) pre[[owneronly, magic(x)]] public view {...}

This starts to get mixed up with invariants though...

[chriseth]

Before we had require, modifiers were used in a very different way. We should collect the current use-cases for modifiers and then maybe restrict the syntax to "pre-conditions, potentially with side-effects" only.

[leonardoalt]

Agree!