nanogui: port question


Previous by date: 7 Jul 2005 10:03:57 +0100 Re: Several fixes to get around compile failures, Alain Volmat
Next by date: 7 Jul 2005 10:03:57 +0100 Re: port question, Ed Sutter
Previous in thread: 7 Jul 2005 10:03:57 +0100 port question, Ed Sutter
Next in thread: 7 Jul 2005 10:03:57 +0100 Re: port question, Ed Sutter

Subject: AW: [nanogui] port question
From: "Alexander Stohr" ####@####.####
Date: 7 Jul 2005 10:03:57 +0100
Message-Id: <000101c582d2$b93f3ec0$9b01a8c0@starlightpc>

RTOS? if you mean a "realtime operating system"
with that then its no problem. it runs on top of
Linux and X11, it has some VNC mode (networking!)
and can be run on several other platfroms. (see
xconfigure for sort of an overview of cross ports)

for better understanding...
nano-X is the grafics driver itselves. it and can
be used with some window manager or even without
a window manager. what you call a "main" function
is actually the main loop of your GUI application.
samples for this sort of coding are in the source
distribution - you might have a look after "nxclock".

realtime? you can set timeouts in the application
and will get some cyclical events - but there is
no gurante that they will arrive timely because
your system might spend time for other duties.
expect some more or less heavy jitter.
further you can set notifiers for IO-events,
such as reception of a character on a serial
line or send buffers running empty on a network
socket. (these are just samples)

for true realtime applications use a system that
has respective extensions like linux with RT patches
or a ready to run RT-distribution. other than that
for most applications it is sufficent to just use
an interrupt driven driver and let the gui application
only get notifications trough device originated events.

lets hope that cleans up a few of the misunderstandings
that i suppose you were currently hazardously involved.

let me say, i would use a nano-X application anytime
where the time accuracy is not that critical, lets say
a computer that has to control some window opener for
a green house on a maximum of 1-sec accuracy or some
weather station data logger - no problem. when it comes
to finer grain, lets say 10ms to 100 ms then it might
work if you dont rely specifically that the selected
time scale is meet relatively often with only minor
disadvantage if the control cycle gets som 5x longer
in rare cases. for high speed regulations loops,
like in engine speed control - you will ned 1 ms or
less *fixed* timing - which you wont adress with a
solely event driven GUI application on base of nano-X.
but nano-X can always be a good help in monitoring
and managing other realtime tasks on your system.

-Alex.

-----Ursprüngliche Nachricht-----
Von: Ed Sutter ####@####.####
Gesendet: Mittwoch, 6. Juli 2005 23:16
An: ####@####.####
Betreff: [nanogui] port question


Hi,
I'm brand new to MicroWindows.  I'm considering using MicroWindows
(i.e. NanoGUI) on a board with no RTOS (for now).  The board has a
framebuffer hooked to an LCD display and runs MicroMonitor as the
bootloader.  As a result, I want to write a simple 'C' main()
function to exercise NanoGUI on this hardware.

Before I go any further, is this outside the focus of NanoGUI or is
it reasonable to think that it can be built for an environment that
is RTOS-less?

Thanks
Ed Sutter



Previous by date: 7 Jul 2005 10:03:57 +0100 Re: Several fixes to get around compile failures, Alain Volmat
Next by date: 7 Jul 2005 10:03:57 +0100 Re: port question, Ed Sutter
Previous in thread: 7 Jul 2005 10:03:57 +0100 port question, Ed Sutter
Next in thread: 7 Jul 2005 10:03:57 +0100 Re: port question, Ed Sutter


Powered by ezmlm-browse 0.20.