Fujimoto Hisa
hisa****@fobj*****
Tue Feb 13 14:22:03 JST 2007
Hi all,
I've merged the bundle/plugin support into the apple-unstable.
there is a known problem, some constants/functions
(NSApp,NSRunAlertPanel)
not works. I can't fix it yet though the point seems RBBundleInit
can't use
around the bridge-support.
thanks
--
hisa
On 2007/02/10, at 12:15, Fujimoto Hisa wrote:
> Hi,
>
> I'm still working on the bundle/plugin support, you can see in the
> bundle-support branch, and it seems almost work good at least about
> SIMBL and VoodooPad plugin, though it has a problem yet (see later). I
> think the api spec may be able to freeze.
>
> I'll merge these into the apple-unstable branch soon (today or a few
> days later).
>
> the API to initialze RubyCocoa and to load a ruby program for a
> bundle/plugin in Objective-C side:
>
> BOOL RBBundleInit(
> const char* path, // relative path to an initial ruby program
> // for the bundle.
> Class klass, // a objc-class binding with the bundle
> id param); // additional parameter for you
> );
>
> An objective-c class, which is defined in a ruby program of the path
> file, bind with the bundle.
>
> And then, the ruby program which is passed by RBBundleInit is like:
>
> require 'osx/cocoa'
>
> OSX.init_for_bundle do
>
> | bundle, # the bundle binding with the klass of RBBundleInit
> param, # the param of RBBundleInit
> logger | # simple logging utility for NSLog
>
> logger.info("bundle=%p param=%p", bundle.to_s, param.to_s)
> require 'my_class'
> ...
> ...
> end
>
> an error raised from the block is rescued by the init_for_bundle
> itself and logging by NSLog, so you have not to care it.
>
> currently two examples eixst:
> sample/VPRubyPluginEnabler - VoodooPad plugin enabler
> sample/RubyAnywhere - SIMBL plugin
>
>
> Well, a problem exists yet. Several constants and functions can not
> access from ruby program loaded by RBBundleInit. theese are e.g.:
>
> NSApp, NSShiftKeyMask, NSRunAlertPanel, etc ...
>
> we should fix these after the merge.
>
> thanks
> --
> hisa
>
> # btw, it may be better to merge the apple-unstable into the trunk
> # already.
>