On Dec 17, 2007 6:13 AM, mario <[EMAIL PROTECTED]> wrote:
> On Dec 16, 2007 11:04 PM, Gustavo Sverzut Barbieri <[EMAIL PROTECTED]> wrote:
> > > I'm still working on my arm board:
> > > All it's ok evas_fb_test run perfectly.
> > >
> > > Now I'm working on edje: there is a problem in a simple program:
> > >
> > Mario, it works great on Nokia devices (arm1136jf-s and arm926ej-s), I
> > never tried with fb engine, just software_16_x11 and software_x11, but
> > I can't see how can this change things.
>
> Solved.
> I used an old (August) version of edje taken from
> http://download.enlightenment.org/snapshots/
> Now I'm using the code from cvs and all it's ok....
>
> ok but slow :(
> better than qtopia but still slow!
> next step: try the software16 engine over kdrive to see if something is 
> better.
> ... even if kdrive seems to be impossibile to compile.

Please, try to write a fb_16 using software_16 as base. I can give you
some hints, but if you know the fb you are working with, just try to
write a dummy engine that does some output, then put software_16 as
the base, look at software_16_x11, the general idea is:
   - create engine info
   - implement drawing functions to be the software_16
   - implement output_redraws_next_update_get() [possible almost the
same!] to get surface that primitives will draw to. See below.
   - implement output_redraws_next_update_push(): Mark somewhere
regions that were dirtied, or send them to display. See below.
   - implement output_flush(): If it's cheaper to send all updates in
one go, send marked regions here. Free resources reserved to draw this
frame.

General idea is quite simple, and to easy your implementation, get an
older software_16_x11 that doesn't provide rotation, you can add it
later. Remember that Evas render each frame, so output_flush() will be
called after each frame. If you want to save allocation-free every
time, you can 1) not free at all or 2) use output_idle_flush() that is
called after some time no render is done (ie: animation ended).   Also
remember that we were dealing with X11, that involves 2 processes and
IPC, then it's cheaper to use shared memory and just mark clip regions
in an array, then inform which regions are dirty in one IPC call...
possible it's the way to go with FB too, since you'll do syscalls in
the end.

> p.s.
> for those who will came: pkg-config fails on cross-compiling:
> sometimes you need to manually enable|disable what you want|dont in
> the configure.in

using what? I just noticed this using scratchbox and running
pkg-config from inside python, i get some "interrupted system calls",
but I think it's more an python-scratchbox issue than something else.


-- 
Gustavo Sverzut Barbieri
--------------------------------------
Jabber: [EMAIL PROTECTED]
   MSN: [EMAIL PROTECTED]
  ICQ#: 17249123
 Skype: gsbarbieri
Mobile: +55 (81) 9927 0010

-------------------------------------------------------------------------
SF.Net email is sponsored by:
Check out the new SourceForge.net Marketplace.
It's the best place to buy or sell services
for just about anything Open Source.
http://ad.doubleclick.net/clk;164216239;13503038;w?http://sf.net/marketplace
_______________________________________________
enlightenment-devel mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/enlightenment-devel

Reply via email to