[haiku-3rdparty-dev] Re: Problem sending/receiving app defined msgs
- From: hudsonco1@xxxxxxx
- To: haiku-3rdparty-dev@xxxxxxxxxxxxx
- Date: Sun, 5 Dec 2010 18:32:03 -0500 (EST)
Guys,
Thanks for taking the time to think about this. I might as well put some cards
on the table. I am working on
another article. This one is on pervasive multithreading and how this makes
some tasks easier to do in Haiku.
In this case an application starts, spawns 5 instances, 4 of them remain on
screen and send messages to each
other. Each application acts as a mind of remote control for any other
application to play an audio file.
The program I am writing is not complicated but shows some interesting
techniques: Using BMessenger,
sending messages between applications, playing audio from a resource, etc. The
real goal is of course to
bring more eyeballs to Haiku and invite more application developers to use
Haiku for interesting projects.
I am a beginner in creating thread safe code, but I want the code to be safe
and to run on any number of
processors. At some point I would like to post the source code and request a
code review.
Here is another question. Is there a bug in BFileGameSound? I used it to play
an audio file from disk. It worked once,
then afterwards doesn't play, gives an error about a bad port. Even after
rebooting.
Thanks,
Andrew
-----Original Message-----
From: Stephan Assmus <superstippi@xxxxxx>
To: haiku-3rdparty-dev <haiku-3rdparty-dev@xxxxxxxxxxxxx>
Sent: Sun, Dec 5, 2010 11:48 am
Subject: [haiku-3rdparty-dev] Re: Problem sending/receiving app defined msgs
Hi,
Von: "Axel Dörfler" <axeld@xxxxxxxxxxxxxxxx>
> Stephan Aßmus <superstippi@xxxxxx> wrote:
> > Am Sonntag, den 05.12.2010, 14:15 +0100 schrieb Axel Dörfler:
> > > Using the window as pointer is indeed not thread-safe - creating a
> > > BMessenger out of it is, though.
> > I remember having traced this code before and I am pretty sure I have
> > found out that invoking PostMessage() on a stale BLooper pointer is
> > save. Making a BMessenger from a stale BLooper pointer boils down to
> > the
> > same code behind the scenes. On the one hand I can't trust my memory
> > too
> > well these days, but on the other hand I am too lazy to trace the
> > code
> > again. So I'll leave that up to you to prove me wrong, if you don't
> > believe me. :-)
>
> I know you are right, but if you have any further questions, you just
> have to read through Ingo's mail who goes into detail of everything,
> and is completely correct, too :-)
Yes I saw that mail and was happy about the insights contained. I just hope
that
when you wrote your initial reply you hadn't seen the mail I already wrote. :-P
Best regards,
-Stephan
Other related posts:
