console,util: implement %o as formatting specifier

[Trott]

• Version: master
• Platform: all
• Subsystem: console,util

console.log() and util.format() support %d, %i, %f, etc. but not %o for objects which seems to be supported in major browsers.

Refs: #10292

/cc @silverwind

[silverwind]

There's also %O which is supported in major browsers. According to the spec it's pretty similar to %o.

%O

An object with generic JavaScript object formatting is a potentially expandable representation of a generic JavaScript object.

%o

An object with optimally useful formatting is an implementation-specific, potentially-interactive representation of an object judged to be maximally useful and informative.

We could pretty-print %O, but other than that, I'm not seeing us implementing more differences. In Browsers, I don't see a difference on them either.

[XadillaX]

So which shall we implement? Or both?

[TimothyGu]

The difference between %o and %O is intended to be much like the difference between console.log and console.dir, with %o being like console.dir and %O being like console.log:

Unfortunately browsers didn't implement %o and %O that way, and in Node.js console.log and console.dir are equivalent anyway. I'd be alright with just util.inspect the objects for both formatters.

[Fishrock123]

Yeah I'd also be fine with it giving our typical object expansion.

@domenic is this about how you'd expect it to work in Node?

[domenic]

Yeah I'd expect either %o or both %o and %O to map to the usual util.inspect. Especially given that there's no opportunity for interaction it makes sense for them to be similar.

If you wanted to come up with some differentiator, you could. E.g., maybe one of them is meant to be more like valid JS syntax, and the other has little annotations like the name of the class or something. Or, %O special-cases certain important Node objects to given them interesting representations beyond just their properties (similar to how %O in browsers on DOM nodes produces an interactive <element> representation). But I'm not sure what those special object representations would be.

But keeping things simple, and having both be the usual object expansion, also makes sense to me.

[Fishrock123]

Marking as good first contribution as I think mapping it would be fairly straightforward.

[gla5001]

is this still available? I'd like to take a stab at it as a first contribution.

[Fishrock123]

@gla5001 it would seem so, yes.

[gla5001]

cool. i'll go ahead and work it.

Just so i'm clear, it seems like the behavior of the '%o' formatting specifier should match the behavior of the '%j' specifier. Is that correct?

[vsemozhetbyt]

@gla5001 I do not think so:

> const obj = { func: function() {} };

> console.log('%j', obj);
{}

>  console.log(util.inspect(obj));
{ func: [Function: func] }

JSON output has restrictions the util.inspect() has not.

[Fishrock123]

@gla5001 No, by the discussion it should just return util.inspect() on the passed value.