[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Re: GNUstep roadmap (was Re: [Suggestion] GNUstep-test for quality contr
Chris B. Vetter
Re: GNUstep roadmap (was Re: [Suggestion] GNUstep-test for quality control)
Mon, 3 Nov 2003 13:52:20 -0800
On Mon, 3 Nov 2003 22:28:16 +0100
Quentin Mathé <address@hidden> wrote:
> Le 3 nov. 03, à 19:02, Chris B. Vetter a écrit :
> > Howl has several problems.
> > One is that it is blocking your application as you need to enter
> > Howl's own runloop. You can try to 'isolate' it in a separate
> > thread, but then again, AppKit itself isn't thread-safe.
> but could we imagine in this case to perform the AppKit calls in the
> main thread (to discard the thread-safety issue)
I'm not sure I understand what you mean.
As far as I digged into Howl, you will have to call their runloop, which
will never return (!) and use some delegate functions that handle
"messages" from the runloop (kinda similar to Apple's implementation).
You can probably "isolate" the runloop in a separate thread (well,
that's what the docu suggests) but, as I said, AppKit isn't thread-safe,
so you might run into several problems here.
> > Another problem is that certain facilities Apple's NSNetService
> > classes depend on aren't implemented, eg. browsing for default
> > domains. There are some more but those two are the first that come
> > to mind.
> ok, but the source are available then it can be added, isn't it ?
Probably, if you're willing to spend the time to figure out how the
stuff in Howl actually works. I just noticed from reading the docu, that
some parts are (or at least seem to be) missing.
Re: GNUstep roadmap (was Re: [Suggestion] GNUstep-test for quality control), Quentin Mathé, 2003/11/02
Re: GNUstep roadmap (was Re: [Suggestion] GNUstep-test for quality control), Chris B. Vetter, 2003/11/03