The other day on IRC I was told that:
Apr 29 00:34:36 <Hoernchen> last thing i heard was that nuttx was
superseded in favor of a more lightweight solution And:
Apr 29 00:41:03 <steve|m>
http://openbsc.osmocom.org/trac/wiki/OsmoDevCon2012/Minutes#a18:30roundtable... which was made back in march(we also started working on nuttx near march) and contains the following:
laf: summary we could not gain much from nuttx, rockbox could provide use
with UI (inspiration)? steve: framebuffer is mosly compatible
I wasn't aware of all that. Also I don't have much details on what was said since I wans't there...
Should me and Alan Carvalho de Assis continue the work we are doing? What is the current plan?
our status is here: http://bb.osmocom.org/trac/wiki/nuttx-bb/drivers
The current work on my side is to: * make serial work without sercomm(done locally,just some configuration change) * unbreak the booting of the calypso(require serial to work without sercomm,the commit that created the problem has been identified, how to fix is a work in progress) * I also tried to change toolchain(I generated a toolchain with openembedded) to see if it fixed the issue but according to Gregory it's not the right fix. * I also identified some dummy functions in the nuttx version of sercomm that we should get rid of.
On Alan Carvalho de Assis side : he's trying to make the keypad work but he has some difficulties with it.
Denis.
Hi Denis,
there was no decision on OsmoDevCon, apart from some opinions.
I'm not aware of anyone actually having reviewed/evaluated rockbox at this point.
The general point was that we apparently don't really _need_ a "full OS" like NuttX, but there are other solutions like rockbox that may bring in other useful parts like a UI for low-resolution dot-matrix displays.
I guess until somebody actually tries and experiments with an early port of rockbox to the calypso, we won't know what is the best way to move forward :/
And in general, everybody seems to be busy with lots of other things, so I'm not sure if (and when) that might happen.
Regards, Harald
Hi, all,
Of course, I am not impartial, but I do want to point out two things. First, there appears to be some misconception that NuttX is big and heavyweight. That is certainly not true. Most implementations use only about 40Kb or so (depending on the features that you incorporate). And you can get it as small as 12Kb; it will run on a z8 or an 8051 or an AVR with 4Kb of RAM. That 40Kb footprint is for a complete environment that might include displays, USB, networking, filesystem, support etc. Well, if you include everything it pushes about 50Kb.
So I have trouble understanding that the issue there. And secondly, NuttX does have a rather complete graphics capability. Probably the best in class:
1. A window system like a tiny X that is scalable from a single thread solution to a full multiple NX server (see http://nuttx.sourceforge.net/NXGraphicsSubsystem.html and http://nuttx.svn.sourceforge.net/viewvc/nuttx/trunk/nuttx/graphics/),
2. An extensive "widget" for adding buttons, textboxes, etc. to NuttX. This has only been released to open source recently but is fairly matured because it has been integrated into commercial products (http://nuttx.sourceforge.net/NxWidgets.html and http://nuttx.svn.sourceforge.net/viewvc/nuttx/trunk/NxWidgets/).
3. Finally, I am currently developing a tiny window manager that I am calling nxwm. This was inspired by the Matchbox Window Manager concept: That is, it is truly multiple window manager but only one window is display in time. This simplification helps performance on LCD based products (in the same way that a tiled window manager helps) and also makes the best use of small displays. It is awkward trying to manage multiple windows on a small display.
nxwm is oriented toward tiny screens with touch capability. It supports a task bar and a start menu with the normal meanings. I have only been working on this for about 3-4 days,and I don't expect to have this complete for another week or so. But it is something else that you should be aware of if graphics support is important to you.
The code is growing here: http://nuttx.svn.sourceforge.net/viewvc/nuttx/trunk/NxWidgets/nxwm/.
But I find that RTOSs tend to be religious topics. People get strong opinions with very little information. So other than providing all of the information that I can and providing the best support that I can, I am not involved in this decision.
Greg
________________________________ From: Harald Welte laforge@gnumonks.org To: Denis 'GNUtoo' Carikli GNUtoo@no-log.org Cc: baseband-devel@lists.osmocom.org; Alan Carvalho de Assis acassis@gmail.com; Gregory N spudarnia@yahoo.com Sent: Monday, April 30, 2012 10:02 AM Subject: Re: should we continue to focus on nuttx?
Hi Denis,
there was no decision on OsmoDevCon, apart from some opinions.
I'm not aware of anyone actually having reviewed/evaluated rockbox at this point.
The general point was that we apparently don't really _need_ a "full OS" like NuttX, but there are other solutions like rockbox that may bring in other useful parts like a UI for low-resolution dot-matrix displays.
I guess until somebody actually tries and experiments with an early port of rockbox to the calypso, we won't know what is the best way to move forward :/
And in general, everybody seems to be busy with lots of other things, so I'm not sure if (and when) that might happen.
Regards, Harald
Hi Gregory,
thanks for your feedback, it is much appreciated.
On Mon, Apr 30, 2012 at 10:14:28AM -0700, Gregory Nutt wrote:
So I have trouble understanding that the issue there. And secondly, NuttX does have a rather complete graphics capability. Probably the best in class:
The question is not whether it has the best graphics capabilities or not. The question is rather: Is it suitable for a 98x67 / 96x64 resolution display like in our target platforms?
To me, the NuttX graphics framework seemed to be aimed at much larger resolution displays, more like a "real" computer with actual windows on a desktop - not like a small cellphone where every pixel counts. Please correct me if that impression was wrong.
So the question was not whether NuttX is good or bad - simply if it is the best possible fit for our specific application :)
Regards, Harald
Hi Harald,
First all sorry top-posting...
NuttX also have support to dot-matrix displays, see this "mp3 player" running NuttX:
http://www.youtube.com/watch?v=A39rsIf07AA
Best Regards,
Alan
On 4/30/12, Harald Welte laforge@gnumonks.org wrote:
Hi Denis,
there was no decision on OsmoDevCon, apart from some opinions.
I'm not aware of anyone actually having reviewed/evaluated rockbox at this point.
The general point was that we apparently don't really _need_ a "full OS" like NuttX, but there are other solutions like rockbox that may bring in other useful parts like a UI for low-resolution dot-matrix displays.
I guess until somebody actually tries and experiments with an early port of rockbox to the calypso, we won't know what is the best way to move forward :/
And in general, everybody seems to be busy with lots of other things, so I'm not sure if (and when) that might happen.
Regards, Harald
--
- Harald Welte laforge@gnumonks.org
http://laforge.gnumonks.org/
"Privacy in residential applications is a desirable marketing option." (ETSI EN 300 175-7 Ch. A6)
First all sorry top-posting...
Hmmm... I did that too.
NuttX also have support to dot-matrix displays, see this "mp3 player" running NuttX:
I'm not certain, but I don't think that this example is using any of the NuttX graphics facilities. I am not sure what he is doing, but the fonts don't look familiar (NuttX supports 17 different Helvetica and Times Roman fonts) and the drawing certainly does not come from Nuttx. I don't have any good demos of the graphics from NuttX but they are all rendered as "3-D" images. I've attached the output of the button array unit test to give you a better idea of what I mean.
(That attached PNG came from here: http://nuttx.svn.sourceforge.net/viewvc/nuttx/trunk/NxWidgets/UnitTests/CBut...)
This button array, by the way, can used with a touchscreen as an on-screen keyboard (with the output going to an edit box). "Widgets" like this have already been used to implement the complete GUI for a medical device.
Greg
baseband-devel@lists.osmocom.org