Re: vfsmm transfer() functions trouble

[Murray Cumming <murrayc murrayc com>]

On Fri, 2005-08-26 at 11:58 +0200, Matthias Kaeppler wrote:
> So, I had a look at the source code of Async::Handle::transfer() and I 
> saw that the result of the C-API call to gnome_vfs_async_xfer() is 
> checked and if it's not equal to GNOME_VFS_OK, an exception is thrown.
> 
> But doesn't that only work if the gnome_vfs_async_xfer() doesn't return 
> until it's finished? Since I suppose that gnome_vfs_async_xfer() runs in 
> an own thread of execution, it's pretty useless to check its result right?

Very possibly. A patch would be welcome.

>   (I also browsed through the code of the C API, but I was not able to 
> figure out what exactly this function does; its body merely consists of 
> a pointer-to-function definition which seems to never be called and a 
> call of some odd macro which I couldn't yet figure out what it does)

The gnome-vfs list would be a good place to ask.

> On top of that, throwing exceptions across thread boundaries is also not 
> possible; I can't catch an exception in one thread which was thrown in 
> another thread.
> 
> On the other hand, since I suppose you know what you're doing, maybe I 
> am missing something? :D

Don't assume that we've understood the C API fully or that we've tested
everything. We need your help.

> I also wondered about the proxy object you're creating which never gets 
> deleted. Why is that needed?

proxy objects provide state information.

However, this seems relevant:
http://bugzilla.gnome.org/show_bug.cgi?id=144627

> Would be nice if you could at least give me some hints... Unfortunately, 
> this newsgroup doesn't seem to be very active/responsive though.

-- 
Murray Cumming
murrayc murrayc com
www.murrayc.com
www.openismus.com