[RFC]Proposal for better garbage collection

[H. S. Teoh]

On Thu, Feb 23, 2012 at 09:18:22PM +0100, Artur Skawina wrote:
> On 02/23/12 20:58, H. S. Teoh wrote:
[...]
> > This is a bit off-topic, but what happens in the current
> > implementation if you pass a D callback to a C function, and then
> > throw an exception from the callback? Does it work? Or does it do
> > something really nasty?
> 
> No, unless you consider a segfault to be really nasty. :)

Well, segfaults are nasty, but there are nastier things. :)


> Actually, it mostly works - i just tried it in a gtk app, and it works
> as long as you catch the exception and only look at the error msg. If
> you don't catch it (or try writeln(e) etc), then the result is
> something like:
> 
> action.Action!(int).Action.registerNS.MissingActionEx at action.d(54): Action "GUI" missing symbol 'int AppWindowClosed()'
> ----------------
> ./gtkapp() [0x8054366]
> /usr/lib/i686/sse2/libgtk-x11-2.0.so.0(+0x153052) [0xf7375052]
> /usr/lib/i686/sse2/libgobject-2.0.so.0(g_closure_invoke+0x19b) [0xf71e25fd]
> /usr/lib/i686/sse2/libgobject-2.0.so.0(+0x1ddc8) [0xf71f2dc8]
> /usr/lib/i686/sse2/libgobject-2.0.so.0(g_signal_emit_valist+0x59a) [0xf71fa37f]
> /usr/lib/i686/sse2/libgobject-2.0.so.0(g_signal_emit+0x34) [0xf71fa61f]
> /usr/lib/i686/sse2/libgtk-x11-2.0.so.0(+0x2a17f3) [0xf74c37f3]
> /usr/lib/i686/sse2/libgtk-x11-2.0.so.0(gtk_main_do_event+0x8e6) [0xf7373856]
> Segmentation fault
> 
> So something appears to get confused while walking the stack; another
> thing to investigate later, i guess...
[...]

Looks like it got confused at the cross-language boundary.


T

-- 
Prosperity breeds contempt, and poverty breeds consent. -- Suck.com