KJSProxy should be removed Basically this is just a big cleanup patch, which removes a bunch of now un-used khtml stuff. We've long sinced forked from KHTML, this is just one of many patches needed to make our code faster and cleaner at the cost of further breaking "compatibility" with khtml.
Created attachment 5077 [details] Proposed patch (just removes code)
Comment on attachment 5077 [details] Proposed patch (just removes code) The patch is unfortunately a bit long, however all it does is remove code so it should be very straightforward to read through. I killed KJSProxy (replacing it with KJSProxyImpl) and removed KLibrary, KGlobal, KInstance, fixed KStandardDirectories::locate to actually be able to load other things besides khtml4.css and quirks.css (means that svg.css will now load). Basically any code that I touched which contained code which should be removed, I removed.
Comment on attachment 5077 [details] Proposed patch (just removes code) Do *not* return interpreterForGlobalObject. That will break LiveConnect and Objective-C bindings. I'll read over the rest of the patch later.
Comment on attachment 5077 [details] Proposed patch (just removes code) Marking review- because removing interpreterForGlobalObject is wrong.
This patch adds SVGEventListener to the project file. Why? Can you get rid of this line too? : settings = view ? view->part()->settings() : 0; It would be nice if you wrote a test for the new locate() function. Or would its failure just break everything?
Hum... I did finally find where interpreterForGlobalObject was used (in kjs/bindings/*). I'll add it back. SVGEventListener is a *very simple* subclass of JSLazyEventListener which I wrote, but decided I would attach to a different patch. I'll make sure to land them near each other. settings = view ? view->part()->settings() : 0; Nope, settings is used later in that class... unless I just pull the settings off of the view/part when their needed, which I could do. Yes, the new locate did break everything (litterally everything) when I got it wrong the first time. Also, now loading svg.css changes the way at least a couple SVG tests are rendered, so I think its covered.
Created attachment 5080 [details] Newer patch (no more SVGEventListener or interpreterForGlobalObject changes)
Comment on attachment 5080 [details] Newer patch (no more SVGEventListener or interpreterForGlobalObject changes) Looks fine. r=me