Almost two weeks after the release of the OS X 10.5.6 update Apple still hasn’t provided a fix for the serious regression with regards to Dvorak keyboard shortcuts in Carbon based applications. Worse they haven’t even indicated that they’re working on it or provided a timeline for a fix or temporary workaround. I did open a bug report with Apple on https://bugreport.apple.com. However as these bug reports are not public it is impossible for others that have been affected by this bug to subscribe to it. Furthermore it arguably makes it difficult for Apple to track how many people have been affected by this issue. Although the thread 10.5.6 breaks Dvorak keyboard shortcuts in Eclipse on the Apple Support Forums makes it clear this issue is causing problems for many people using just as many different applications.
With Apple not taking their responsibilities I figured I’d give it another try to find a workaround. Problem is that I am no OS X developer and though I do know the basic difference between Carbon and Cocoa applications I know next to nothing on OS X application development. So all I have to go on is my intuition and hints I find on the Web.
With some of the keyboard shortcuts working (cmd-s) and others not (cmd-c, cmd-v) I still suspect the culprit to be with an incorrect keyboard mapping instead of some logic error in a piece framework code. I my previous post I created a new Dvorak mapping and installed it. This didn’t solve the problem as it didn’t change any of the mappings specified in the Carbon Framework. At least that’s what I suspect the reason to be.
So is there a way to adjust the mappings in the Carbon Framework? It seems there is. Hinted by references to HIToolbox in the Eclipse bug report on this issue, and a path to this part of the Carbon Framework mentioned on http://wordherd.com/keyboards/ I figured I had to replace the Dvorak mapping in:
Using ResKnife, an open source resource editor, I had a look in Localized.rsrc and did indeed found a Dvorak KCHR resource (with ID 16300). With the form on http://wordherd.com/keyboards/ I created a layout.r using the following settings:
In the resulting layout.r file I changed the type uchr to KCHR and its ID to 16300. Then I made a backup copy of Localized.rsrc and executed:
gkoller@Kardamom (593) $ /Developer/Tools/Rez -append -useDF -o \ /System/Library/Frameworks/Carbon.framework/Frameworks\ /HIToolbox.framework/Resources/English.lproj/Localized.rsrc \ layout.r
For all I know my attempts to find a workaround are completely misdirected and naive, but I just had to try as this issue is driving me nuts. Maybe this post might give others a few hints and put them on the right track to find a solution, or more likely prevent them from wasting time on similar futile attempt. However it would be much better for Apple to step in and take their responsibility in fixing this regression!