"No power. So I rewired it!" -- Tim Allen
Please bang away on the demo and give me your feedback. Find bugs in the demo? Comment to the main bug. Think the UI stinks? Comment to the main bug. Have other ideas for features? Comment to the main bug.
If you just want a direct link, try out the new, improved, mostly-working DOM Inspector JavaScript Object Viewer.
One neat feature is that, at least for now, you can manually reset the subject of the viewer (what the target is). Try this in your Mozilla location bar with the demo loaded:
javascript:viewer.subject = window;
I know I've asked if eval() is evil() in chrome before, but in the context of DOM Inspector and this object viewer, I don't think that matters too much. You have to be pretty knowledgeable to trick someone into breaking their computer with this.
On a side note, I think this might be powerful enough to be a standalone chrome application. If enough people like it, I'd probably fork it off to a chrome://objectviewer/content/ URI, and have DOM Inspector use it as an overlay. (Although even I have to admit I'd have to see a lot of demand from the community for mozilla.org to justify it.)
Then run framesetOuter.xul from chrome, and see what happens.
#document node now appears as root node in DOM Nodes Viewer.
I am a happy man. That was one of very few patches I had basically written on my own for Inspector which didn't need localization.
I may have to bump several others needing localization back to 1.6alpha. The new features aren't necessarily stuff you want in 1.5beta.
Latest attempt at fixing bug 193942
Bug 193942, JavaScript Object Panel needs more features
I still can't figure out a good way to determine if an object has unenumerated numbered properties...
I'll let Christopher Aillon (module owner for Inspector) speak on long-term planning for Inspector, but I felt I should just take a moment to summarize just what I've been up to for Inspector.
First off, if you really want to look at all bugs filed against DOM Inspector, here's a Bugzilla query:
http://bugzilla.mozilla.org/buglist.cgi?query_format=&short_desc_type=allwordssubstr&short_desc=&product=Browser&component=DOM+Inspector&long_desc_type=substring&long_desc=&bug_file_loc_type=allwordssubstr&bug_file_loc=&status_whiteboard_type=allwordssubstr&status_whiteboard=&keywords_type=allwords&keywords=&bug_status=NEW&bug_status=ASSIGNED&bug_status=REOPENED&emailassigned_to1=1&emailtype1=exact&email1=&emailassigned_to2=1&emailreporter2=1&emailqa_contact2=1&emailtype2=exact&email2=&bugidtype=include&bug_id=&votes=&changedin=&chfieldfrom=&chfieldto=Now&chfieldvalue=&cmdtype=doit
A lot of those bugs catch my eye. Among them is bug 109682, my ultimate goal in the long-term: making Inspector into a fifth panel on Editor.
Bug 112775 is my first major change (by major, I mean something which requires significant tinkering to figure out how to do it right), for something that you'd think would be simple: inserting a node. Turns out there's a lot of possibilities to consider.
Bug 112922 is a bug for editing and splitting text nodes -- a pretty important feature of a DOM editor tool. I haven't yet explained my approach, so in a nutshell, I'm thinking of textareas which update the node as you type. This is consistent with how Editor works, and it's seamless (WYSIWYG).
It also makes handling undo and redo operations in Inspector a bit touchy. Bug 179621 is a multi-phase process for converting Inspector to use Editor's Transaction Manager model, which governs undo/redo for the major applications of Mozilla (Navigator, Composer, Mail/News).
The bread and butter of DOM Inspector are the various viewers -- the applets that let you see the structure of the document and interact with said document. Big changes are coming. I already mentioned bug 112775 (that affects the DOM Nodes, or dom.xul, panel) and bug 112922 (that affects the DOM Node, or domNode.xul, panel). Looking from viewer to viewer, there are other improvements to note.
In dom.xul, we're finally going to see the #document node. Bug 156072. We're also going to see other #document nodes which are "contentDocument" properties of framed documents (like the HTML iframe's contents). That one is bug 201585, and bzbarsky is handling it. (Sorry, I can't hack C++.)
As a direct result of bug 201585, we'll be able to look at stylesheets for framed documents as well. (Bug 201586.) That'll be the Stylesheets viewer for a document, stylesheets.xul.
For specific DOM nodes selected in the left panel, the right panel will see a lot of new information, tailored to newly-exposed nodes such as document nodes, document-type nodes, and processing instructions. Bug 201129.
I'm not touching the box model. Fuggedaboudit.
XBL bindings, I should probably take the time to redesign it a little bit. The interface is confusing. I'm not going to file a bug on that yet, or accept one (if you readers get any ideas). but it's itching in the back of my mind.
Style Rules, well, we need a way to add a new style rule. I'm waiting to hear back from Mr. Aillon as to whether his bug on changing how Inspector accesses style rules includes that or not. If not, I'll file then.
Bug 192841, for non-CSSStyleRule rules, is pretty important to fix for this panel.
Computed Style: It's a simple interface, doesn't need anything from me.
JavaScript Object: the most drastic changes of all the panels, by far. You won't recognize it at first. Bug 193942. New features:
* Dynamic resorting by alphabetical listing or by original ordering
* Revealing unenumerated properties (like window.Components)
* Regular-expression searches of the tree for names or values matching a regexp
* A multiline textbox for those multiline values
* Typeof included
* Matching an inspected value against a higher inspected value (for instance, document.defaultView.document == document)
* Assigning a match for convenience
* Creating new properties and methods on the fly
* Executing methods on the fly and getting a return value added to the tree
* I think that's all I've come up with so far...
Other bugs:
I really should take the time to fix bug 111411.
I filed bug 119388, and I should investigate that one further.
Bug 121774, I should ask for reviews on for what I've got, and again I wait for Mr. Aillon for style rules.
Bug 123089 sounds interesting.
Bug 128421 I think bzbarsky is going to take.
Bug 150560 is another I should take the time to fix.
Bug 158826 will almost certainly need a DOM extension if it's ever going to happen.
Bug 166749, shouldn't be too hard.
Bug 178582 and bug 178583, I'm not sure why I filed those...
Bug 183060 I will probably spend some time fixing. Seems easy.
Bug 189950 I came up with a fix for, and it's the wrong one...
Bug 193724 is assertion-cleanup (so I've heard, since I don't work with assertions) and minor UI improvements.
That's the news so far. We're looking at a timeline for most of the bugs (except under "Other Bugs") of roughly Mozilla 1.5 or 1.6 beta.
Started doing some tinkering around for bug 179621. Here's a documentation/logic breakdown for the first phase patch.
Is anybody supposed to super-review patches for DOM Inspector or not?
Bug 193726 has a simple patch, good docs in the bug, and it's been sitting for over a month.
It's blocking me on several other bugs for DOM Inspector. Including bug 112775, a very largish checkin that I'd like to get in for 1.4alpha. I don't dare seek reviews on 112775 without 193726 fixed.
UPDATE: Thanks to roc for sr= on 193726. I get to eat crow, because caillon meant Venkman didn't necessarily need sr=, but Inspector does.
Just put up the seventh proposed patch to bug 112775 @ bugzilla.mozilla.org -- the primary "create nodes" bug for DOM Inspector. Works fine, beautifully, no JS errors or strict warnings I'm aware of. Only problem is you can cause Mozilla to hang by following a certain sequence of steps.
Which sucks vacuum.
I dunno. All I can say is that there's no way in hell my code's responsible for the hang, and I can prove that.