Bug 73078 – Web Inspector: refactor resource tree model and introduce explicit class for ResourceTreeFrame.

• Status: RESOLVED
• Resolution: FIXED
• Component: Web Inspector (Deprecated)

[Pavel Feldman]

Otherwise we are managing frames and their resources using too many maps.

[Pavel Feldman]

Created attachment 116504 [details]
Patch

[Timothy Hatcher]

Comment on attachment 116504 [details]
Patch

View in context: https://bugs.webkit.org/attachment.cgi?id=116504&action=review

> Source/WebCore/inspector/Inspector.json:680
> +                    { "name": "frameId", "$ref": "FrameId", "description": "Frame identifier.", "hidden": true },
> +                    { "name": "loaderId", "$ref": "LoaderId", "description": "Loader identifier." },

These should be added to the end as optional right? Since this is a change from Inspector-0.1.json. Also why is frameId always hidden?

> Source/WebCore/inspector/front-end/ApplicationCacheModel.js:59
> +        var frame = /** @type {WebInspector.ResourceTreeFrame} */ event.data;

These type comments are getting intrusive…

[Pavel Feldman]

(In reply to comment #2)
> (From update of attachment 116504 [details])
> View in context: https://bugs.webkit.org/attachment.cgi?id=116504&action=review
> 
> > Source/WebCore/inspector/Inspector.json:680
> > +                    { "name": "frameId", "$ref": "FrameId", "description": "Frame identifier.", "hidden": true },
> > +                    { "name": "loaderId", "$ref": "LoaderId", "description": "Loader identifier." },
> 
> These should be added to the end as optional right? Since this is a change from Inspector-0.1.json. Also why is frameId always hidden?

We can add any parameters into the _events_ and remove optional parameters from them. We can't add non-optional parameters to the _commands_ only.

> 
> > Source/WebCore/inspector/front-end/ApplicationCacheModel.js:59
> > +        var frame = /** @type {WebInspector.ResourceTreeFrame} */ event.data;
> 
> These type comments are getting intrusive…

I know. We are experimenting with this type of cast. It is not clear how much we need to annotate to get a strict type checking. But as is, compilation already catches a handful of bugs for me every day.

[Timothy Hatcher]

Comment on attachment 116504 [details]
Patch

View in context: https://bugs.webkit.org/attachment.cgi?id=116504&action=review

>>> Source/WebCore/inspector/front-end/ApplicationCacheModel.js:59
>>> +        var frame = /** @type {WebInspector.ResourceTreeFrame} */ event.data;
>> 
>> These type comments are getting intrusive…
> 
> I know. We are experimenting with this type of cast. It is not clear how much we need to annotate to get a strict type checking. But as is, compilation already catches a handful of bugs for me every day.

I'm confused. Why does order not matter for events when other clients might be using this and break after this change?

[Timothy Hatcher]

(In reply to comment #4)
> (From update of attachment 116504 [details])
> View in context: https://bugs.webkit.org/attachment.cgi?id=116504&action=review
> 
> >>> Source/WebCore/inspector/front-end/ApplicationCacheModel.js:59
> >>> +        var frame = /** @type {WebInspector.ResourceTreeFrame} */ event.data;
> >> 
> >> These type comments are getting intrusive…
> > 
> > I know. We are experimenting with this type of cast. It is not clear how much we need to annotate to get a strict type checking. But as is, compilation already catches a handful of bugs for me every day.
> 
> I'm confused. Why does order not matter for events when other clients might be using this and break after this change?

Sigh, that is not the comment I replied to.

[Vsevolod Vlasov]

Comment on attachment 116504 [details]
Patch

View in context: https://bugs.webkit.org/attachment.cgi?id=116504&action=review

> Source/WebCore/inspector/front-end/ResourceTreeModel.js:108
> +                this._frameDetached(this.mainFrame.id);

Please move this to _frameNavigated.

> Source/WebCore/inspector/front-end/ResourceTreeModel.js:117
> +    _frameNavigated: function(framePayload)

You should remove all child frames here.

[Pavel Feldman]

Created attachment 116524 [details]
[Patch] First pass of review comments addressed

[Yury Semikhatsky]

Comment on attachment 116524 [details]
[Patch] First pass of review comments addressed

View in context: https://bugs.webkit.org/attachment.cgi?id=116524&action=review

> Source/WebCore/inspector/front-end/ResourceTreeModel.js:154
> +        if (!framePayload.parentId && !WebInspector.networkLog.resources.length) {

Please comment on this logic.

[Pavel Feldman]

Committed r101154: <http://trac.webkit.org/changeset/101154>