904739 - rename actors' grip/form methods to toJSON

Status: RESOLVED DUPLICATE of bug 698841

(DevTools :: Debugger, defect, P3)

Description

[Nick Fitzgerald [:fitzgen] [⏰PST; UTC-8]]

This is something that it seems like everyone who works on the debugger server gets confused about. No one knows what a grip or a form is, or how they are different (which we also mess up fairly frequently). It would be a lot more obvious if we just rename them all toJSON, and then we can get the magical JSON.stringify'ing auto-calling toJSON as an added bonus (no more thing.grip() all over the place when sending packets, just use thing! woo!).

Comment 1

[Anup Kumar [:reznord]]

Hi,

I am willing to work on this bug. So, I request you to assign me to this bug.


Thanks in Advance,

Regards,
A. Anup

Comment 2

[Nick Fitzgerald [:fitzgen] [⏰PST; UTC-8]]

Anup, I don't think this is a good bug to work on just yet, I expect there to be some discussion and decisions to be made first.

Take a look at one of these instead: https://bugzilla.mozilla.org/buglist.cgi?j_top=OR&f1=component&list_id=7519021&o1=allwords&resolution=---&emailtype1=exact&status_whiteboard_type=allwordssubstr&emailassigned_to1=1&query_format=advanced&status_whiteboard=mentor%3D&bug_status=UNCONFIRMED&bug_status=NEW&bug_status=REOPENED&bug_status=VERIFIED&email1=nobody%40mozilla.org&v1=Developer%20Tools&product=Firefox

Comment 3

[Jim Blandy :jimb]

I like the toJSON idea a lot. The actors would just get toJSON methods, and then we'd stick them directly in the packets, right?

But we need a precise, quick term that means what "grip" does now. Where the protocol spec says things like:

"This represents an environment introduced by a <tt>with</tt> statement whose operand is <i>object</i> (a grip)."

how would we write that sentence now, in a way that *spells out* what <i>object</i> must be? The spec is not the place to say, "ah, everyone knows that I really mean a grip when I say 'value'," right?

Comment 4

[Jim Blandy :jimb]

(In reply to Jim Blandy :jimb from comment #3)
> I like the toJSON idea a lot. The actors would just get toJSON methods, and
> then we'd stick them directly in the packets, right?

... except that we don't really use JSON.stringify in LocalDebuggerTransport or ChildDebuggerTransport.

Comment 5

[Brandon Benvie [:benvie]]

(In reply to Jim Blandy :jimb from comment #4)
> (In reply to Jim Blandy :jimb from comment #3)
> > I like the toJSON idea a lot. The actors would just get toJSON methods, and
> > then we'd stick them directly in the packets, right?
> 
> ... except that we don't really use JSON.stringify in LocalDebuggerTransport
> or ChildDebuggerTransport.

And hopefully we don't run into any more bugs because of the differences between freezing and JSON encoding.

> freeze({ foo: Infinity, bar: -0 })
> { foo: Infinity, bar: 0 }

> remotify({ foo: Infinity }, bar: -0 })
> { foo: { type: "Infinity" }, bar: { type: "-0" } }

Comment 6

[Nick Fitzgerald [:fitzgen] [⏰PST; UTC-8]]

(In reply to Jim Blandy :jimb from comment #4)
> (In reply to Jim Blandy :jimb from comment #3)
> > I like the toJSON idea a lot. The actors would just get toJSON methods, and
> > then we'd stick them directly in the packets, right?
> 
> ... except that we don't really use JSON.stringify in LocalDebuggerTransport
> or ChildDebuggerTransport.

D'oh, you are correct. I still think toJSON is a better name event without the nice magic.

Comment 7

[Jim Blandy :jimb]

Nick, what would you do about comment 3?

Comment 8

[Brandon Benvie [:benvie]]

Bikeshed the name! The concept is sound (normalizing how we name this stuff and make it more self-explanatory).

toRemote
remotify
remotable
transportable

Or even toGrip, along with a detailed explanation of what a "grip" is. At the very least, the name "form" needs to go because it has no explanatory power at all.

Comment 9

[Nick Fitzgerald [:fitzgen] [⏰PST; UTC-8]]

remotify is pretty good: "... whose operand is object (remotified)"

I also like plain old "serialize": "... whose operand is object (serialized)."

Comment 10

[Rob Campbell [:rc] (:robcee)]

+1 for serialize.

(I was going to suggest "pickle" just to introduce a pythonism)

Comment 11

[Nick Fitzgerald [:fitzgen] [⏰PST; UTC-8]]

This should just go away...