Two independent, but related, issues: 1) cmd-[ on mac will navigate the parent page in addition to changing the inspector panel. All page state is lost as the page is navigated. 2) On a french keyboard, to type a ], you use `alt+gr + )`. Typed in the console, this trigger the inspector to go to the Elements panel. Interestingly [ causes no problems. This makes it impossible to type an array literal in the console.
*** Bug 86955 has been marked as a duplicate of this bug. ***
(In reply to comment #0) > Two independent, but related, issues: > > 1) cmd-[ on mac will navigate the parent page in addition to changing the inspector panel. All page state is lost as the page is navigated. I'm unable to reproduce this one on a recent build (May 11). > 2) On a french keyboard, to type a ], you use `alt+gr + )`. Typed in the console, this trigger the inspector to go to the Elements panel. Interestingly [ causes no problems. This makes it impossible to type an array literal in the console. Assuming this one conveys the contents of bug 86955, will submit a patch for this one shortly.
Darin, I vaguely remember you hack on the keyboard event handling in Chromium. Do you have a clue if the following behavior is correct or not? Chromium produces exactly the same Event data for "Ctrl"+"Alt"+"[" on the English QUERTY keyboard and for "AltGr"+")" (to enter the "]" character) on the French AZERTY keyboard (yes, in the latter case the event flags are: altGraphKey: false altKey: true ctrlKey: true keyCode: 219 keyIdentifier: "U+00DB" metaKey: false shiftKey: false ). Or should we monitor the modifier key states manually...
(In reply to comment #2) > > 1) cmd-[ on mac will navigate the parent page in addition to changing the inspector panel. All page state is lost as the page is navigated. > > I'm unable to reproduce this one on a recent build (May 11). I can repro in Canary and a recent WebKit nightly. Steps: 1) on this page, click to bug 86955 2) cmd-opt-i to open inspector (mine currently opens to Scripts) 3) cmd-[ Results: Inspector moves to Network while parent page navigates back to this ticket
(In reply to comment #4) > (In reply to comment #2) > > > 1) cmd-[ on mac will navigate the parent page in addition to changing the inspector panel. All page state is lost as the page is navigated. > > > > I'm unable to reproduce this one on a recent build (May 11). > > > I can repro in Canary and a recent WebKit nightly. > > Steps: > > 1) on this page, click to bug 86955 > 2) cmd-opt-i to open inspector (mine currently opens to Scripts) > 3) cmd-[ > > Results: > > Inspector moves to Network while parent page navigates back to this ticket OK, I get the point. This works as intended, since on Mac, Cmd-[ is a standard Chrome shortcut to navigate back to the previous page (and Cmd-] navigates forward in the history). This works the same way as Alt-Left/Right on Win and Linux.
(In reply to comment #5) > OK, I get the point. This works as intended, since on Mac, Cmd-[ is a standard Chrome shortcut to navigate back to the previous page (and Cmd-] navigates forward in the history). This works the same way as Alt-Left/Right on Win and Linux. Oops, we seem to intercept this shortcut, but not quite...
Created attachment 144040 [details] Patch
Comment on attachment 144040 [details] Patch View in context: https://bugs.webkit.org/attachment.cgi?id=144040&action=review > Source/WebCore/inspector/front-end/InspectorView.js:46 > + this._filteredKeyIdentifiers = ["U+005B", "U+00DB", "U+005D", "U+00DD"].keySet(); Let's create this._leftBracketKeyIdentifiers and this._rightBracketKeyIdentifiers fields and use them here and in _keyDownInternal to avoid duplicating this confusing unicode magic. > Source/WebCore/inspector/front-end/InspectorView.js:95 > + if (event.charCode === 91 || event.charCode === 93) { Please add a comment saying this is checking for brackets or use some function to convert character to ascii code. > Source/WebCore/inspector/front-end/InspectorView.js:103 > + if (!WebInspector.isWin() || !this._filteredKeyIdentifiers.hasOwnProperty(event.keyIdentifier)) { Please add a comment explaining this workaround (and reasoning) here.
Committed r118520: <http://trac.webkit.org/changeset/118520>
*** Bug 84092 has been marked as a duplicate of this bug. ***