On 1/1/06, [EMAIL PROTECTED] <[EMAIL PROTECTED]> wrote: > Well, before going further, I would like to replace some things. > In fact what I would like to do is to port the libxfcegui4 to directfb. This > library implement some of > the standart we can found on Freedesktop. > (http://www.freedesktop.org/wiki/Standards). The problem is that these > standart are meant to be used in a X11 Environnement. Or, DirectFB does not > always offer the same functionnality. The problem of properties for each > window is just one case among other. (There is also problems with events > because the X11 and DFBevent are not always compatible). > Now, to implement all that, I see 2 possibilities: > > 1) Use our own standarts. This means to provide the same functionnalities > that all who is debated on Freedesktop.org by using different protocols, so > that they can be implemented with DirectFB. The good point is the fact that > it will save work for the DirectFB team. On the other side it will require > more work to port applications, and new DirectFB-specific standarts will have > to be defined and maintened. > > 2)We can keep the current standarts and implement them on the top of > DirectFB. This require to have what I will call a compatible API inside > DirectFB. By this, I mean that the API must be able to deliver the same > functionnality that the Xlib does. This is why I asked if there was a way to > attach named properties on each windows and share them between application. > As I said above, this is not the only things who will be required. (Perhaps > dressing a list of the things we need can be useful). >
That would be very useful !! I'm not sure that we can really address the issue without the list. > Now about Hessian. After having a look, I'm not sure implementing Hessian is > really required for 2 reason: > -This protocol doesn't seem to be "compatible" with the communication made > through X11, so it doesn't really helps in that sense > -I think fusion is able to handle most (all?) of the things we need. > Yes your right I've come to the same conclusions. > But Mike has far more experience than me on that kind of things (in fact i'm > just new on these subjects, and I'm still trying to understand all I need to > do),and also more interest. (I do it for fun, I don't think my company will > give me a day anything for that). > Thanks but not really porting/emulation of existing api's is always hard. In many cases GTK has simply wrapped a number of X11 api's its not clear that all of the wrappers are really needed to write GTK programs. I'd say we concentrate on the core functionality and not perfect emulation of the wrappers. Gdk is for example far more complex then it needs to be to provide basic support for a widget library. It's not clear that a perfect implementation of gdk is needed or desired for DirectFB what is important is that a GTK app for directfb should compile and run with little porting on the X11 implementation not the reverse. On the GTK time I think this series of releases should probably be the end of life for the 2.x series and we should move on to a 3.x series that eliminates older api's and consolidates and simplifies the gdk layer. A relatively popular DirectFB gdk port could help identify the really important api's. I have spent some time with the fusion api and am now confident I can implement the needed fusion bindings once its clear what they should be. Mike > > > On Sat, 31 Dec 2005 10:31:22 -0600 > Mike Emmel <[EMAIL PROTECTED]> wrote: > > > Okay I went through fusion in detail. > > I don't see a reason why you can layer hessian on top but on the same > > hand I'm not sure its required right now. > > I've played with fusion a bit and I'm pretty happy with it. > > Only one question. > > > > Right now the way to get the fusion world is via knowing its index > > which is stored in a enviroment variable or given on the command line. > > I assume this is pretty much the only way so on the case of me adding > > a new shared world for application we simply need a new environment > > variable. > > Or we can easily connect intially to the main world to get the index > > of a secondary shared world. It would still need to be created in the > > master process ? > > > > Mike > > > > > > On 12/31/05, Mike Emmel <[EMAIL PROTECTED]> wrote: > > > On 12/31/05, Denis Oliver Kropp <[EMAIL PROTECTED]> wrote: > > > > [EMAIL PROTECTED] wrote: > > > > > On Fri, 30 Dec 2005 21:45:30 +0100 > > > > > Denis Oliver Kropp <[EMAIL PROTECTED]> wrote: > > > > > > > > > > > > > > >>Mike Emmel wrote: > > > > >> > > > > >>>On 12/30/05, Ben Combee <[EMAIL PROTECTED]> wrote: > > > > >>> > > > > >>> > > > > >>>>DirectFB already uses Fusion for shared memory access and IPC, so > > > > >>>>it's > > > > >>>>probably something you should consider. > > > > >>>> > > > > >>> > > > > >>> > > > > >>>I'm a bit concerned about sending application level messaging via > > > > >>>fusion. > > > > >>>Generally its for fairly critical messages. There is little control > > > > >>>on how much messaging the applications will do. > > > > >> > > > > >>We can have a separate Fusion World for Application Level IPC. > > > > >> > > > > >>It would use another Fusion Device, Dispatcher Thread and SHM Pool. > > > > >> > > > > >>-- > > > > >>Best regards, > > > > >> Denis Oliver Kropp > > > > >> > > > > > > > > > > A question here can be how is unique intented to work and how will > > > > > unique read/write properties (like the fact that an application gets > > > > > raised or lowered, or the fact that an application sets an emergency > > > > > hints because it reclams the attention of the user, wich generally > > > > > result in an apps changing his color in the taskbar,...) and how it > > > > > will communicate with the other apps. > > > > > In the X11 world, this kind of thing is done by getting/setting > > > > > (named) properties of each window. Is it possible to do the same kind > > > > > of things with fusion? (and how?) > > > > > > > > This basic mechanism can be added to the CoreWindow, but is that all you > > > > need? > > > > > > > > > > Thats pretty close to all X11 does for Atoms but you also need to > > > support something like XClientMessageEvent. > > > > > > I'd really like to just implement this > > > http://www.caucho.com/resin-3.0/protocols/hessian-1.0-spec.xtp > > > > > > On top of fusion for the transport. > > > Its almost the smallest generic serialization/messaging protocal you > > > can have. I'm pretty confident we can build arbitray messaging api's > > > on top. > > > > > > For example we could even support the full X11 protocol with a message > > > converter listening on the wire. > > > > > > Mike > > > > > > > -- > > > > Best regards, > > > > Denis Oliver Kropp > > > > > > > > .------------------------------------------. > > > > | DirectFB - Hardware accelerated graphics | > > > > | http://www.directfb.org/ | > > > > "------------------------------------------" > > > > > > > > _______________________________________________ > > > > directfb-dev mailing list > > > > [email protected] > > > > http://mail.directfb.org/cgi-bin/mailman/listinfo/directfb-dev > > > > > > > > > _______________________________________________ > directfb-dev mailing list > [email protected] > http://mail.directfb.org/cgi-bin/mailman/listinfo/directfb-dev > _______________________________________________ directfb-dev mailing list [email protected] http://mail.directfb.org/cgi-bin/mailman/listinfo/directfb-dev
