There are still more features I would like to implement, and the direct bitmap
rendering is not there yet, but PGFX works in immediate mode or persistent mode now. I
added demos for both to the cli_c/demos directory.
While adding the PG_GROPF_TRANSIENT flag to the server to support the PGFX_IMMEDIATE
rendering mode, I also improved the memory management for gropnodes. When a gropnode
is freed, it is moved to a 'zombie gropnode' list. when a new grop is needed, one is
taken from this zombie list. This eliminates memory allocation and deallocation during
normal drawing operations.
This should be enough to get the basic graphical output working nicely, and because
the pgfx backends can be switched easily, switching to a bitmap-based rendering engine
will require very few changes in your code.
On Fri, 13 April 2001, Micah Dowty wrote:
>
> On Fri, 13 April 2001, Olivier Bornet wrote:
>
> >
> > Hi Micah,
> >
> > > Just to let everyone know where they are in the queue :)
> > > Here's my current short-term todo list:
> > > 1. Write PGFX
> > > 2. Tomorrow (friday) night fix ez328 rotate bug
> > > 3. Re-test and fix ncursesinput
> > > 4. Make video drivers fast
> >
> > Looks good for me. And if it's written by priority, very great : the
> > PGFX will be available for the 1st test very soon. I'm very happy, and
> > impatient to test.
> >
> > > The PGFX interface functions themselves are nice short primitives,
> > > that operate on a pgcontext. The pgcontext contains a table of function
> > > pointers that implement the primitives, and it contains information
> > > about the rendering device.
> >
> > The pgcontext structure seems fine for me. I'm just curious of what is
> > the device field. Is it just for example the canvas used when we create
> > the pgcontext with pgnewCanvasContext ?
>
> Well, the apps probably shouldn't be doing anything to the pgcontext anyway because
>it's implementation dependant. But, yes, the device field can be a canvas widget. If
>the context is outputting to a bitmap, it might be a bitmap handle.
>
> >
> > > The app will obtain a pgcontext using a rendering-device-specific
> > > function, then render to it without regard to what it actually is (well,
> > > mostly). I have started a pgfx backend for the canvas widget. There will
> > > be an 'immediate' mode and a 'persistent' mode. The immediate mode will
> > > implement a more traditional drawing model where each group of
> > > primitives is packaged, sent to the server, and rendered, but not
> > > stored. The persistent mode would expose more of PicoGUI's gropnode
> > > manipulation functionality and allow the server to do more work for the
> > > client, at the cost of some complexity in the client.
> >
> > Great ! The immediate mode is just what I need now for using PicoGUI
> > with the current Waba implementation.
> >
> > > When the bitmap-based primitives in pgserver are available, a pgfx
> > > backend will be written for them allowing the app to easily switch
> > > rendering methods. (enable/disable double-buffer for example)
> >
> > Seem also OK. I think I can use this for loading bitmap file (.bmp
> > Window$ files) when this will be available.
>
> Well, not really. This will let apps write into bitmaps but it's not really intended
>for loading purposes. That should be done in the server.
> I was planning on implementing a BMP loader in the server soon, when do you need
>this?
>
> >
> > > I would suggest anyone interested in the client-side graphics
> > > operations take a glance through pgfx.h and make sure it's reasonable,
> > > because that's the API you'll be using :)
> >
> > The API looks good for me. Thanks a lot.
> >
> > Thanks for this good jobs, and have a nice day.
>
> Happy Friday!
> :^D
>
> >
> > Olivier
> > --
> > Olivier Bornet SMARTDATA SA
> > [EMAIL PROTECTED] Centre du Parc
> > http://www.smartdata.ch av. des Pr�s-Beudin 20
> > Phone +41-27-723'55'03 1920 Martigny
> > Fax +41-27-723'55'19 Phone +41-27-723'55'18
> >
> > _______________________________________________
> > Pgui-devel mailing list
> > [EMAIL PROTECTED]
> > http://lists.sourceforge.net/lists/listinfo/pgui-devel
>
> --
> perl -e'socket S,2,1,tcp;connect S,pack(H16,"0200029a80692d32").\0x8;print
> grep!/=/,<S>'
>
>
> _______________________________________________
> Pgui-devel mailing list
> [EMAIL PROTECTED]
> http://lists.sourceforge.net/lists/listinfo/pgui-devel
--
perl -e'socket S,2,1,tcp;connect S,pack(H16,"0200029a80692d32").\0x8;print
grep!/=/,<S>'
_______________________________________________
Pgui-devel mailing list
[EMAIL PROTECTED]
http://lists.sourceforge.net/lists/listinfo/pgui-devel
