Release Why not Maemo?

[ChairmanNow]

(In case you don't know what Maemo is, it is the Linux-based OS for the Nokia N770/N800/N810 handhelds - see here)

I've been following the thread regarding Ubuntu, and it seems to me that a lot of folks are very keen to re-invent the wheel. Why not use a really good wheel that currently exists? That'd be Maemo.

Some advantages of adapting/forking Maemo for use on the Pandora:

+ A distribution that's been optimized for use on a device with very similar specs and form-factor to the Pandora
+ Ability to re-use all of the apps that have been ported and optimized for a device with the same screen resolution
+ Existing packaged dev tools
+ Ability for commercially minded developers to offer their software to a wider audience

Nokia has invested a huge amount of time and money creating a distribution that's optimized for performance on an ARM based handheld with the same amount of memory, and has produced an UI that's likewise optimized for the resolution that the Pandora uses. Why not take advantage?

I'm kind of a newbie to this community, so this may have already been discussed and dismissed. I'd sure like to know why if that's the case.

 

[jchord]

Well some parts of it will need to be rewritten as parts of it is closed source. And drivers will have to be rewriten. Otherwise I think it would be cool. It would be nice to join the nokia internet tablet community and the pandora community through using the same os. It would benefit both devices.

edit: I know I would love to have the hava player on a pandora.

 

[craigix]

It does run but it isn't all open source.

 

[Tor]

It could be an interesting exercise. The devices are quite similar. The N800 Nokia model has wi-fi, bluetooth, 800x480 screen, dual sd slots, 128MB RAM, 256MB NAND, but no keyboard. The N810 has a keyboard and there are some other differences. And there is a camera in both but who cares.

I imagine that you should start with a Pandora-specific kernel, to handle the (presumably different) wifi and bt chipsets. Then get a Pandora-specific X server, because of the different graphics chipsets. Then the rest of the actual maemo should go in more or less unmodified. What I don't know is if building an X server for Pandora (an armel version) is possible, presumably there's a closed-source driver for the Power VR support, or maybe there is a pre-compiled one available already? The maemo distribution is basically a Debian-based armel distribution.

So: Sort out kernel/drivers and X-server (graphic drivers), the rest should be quite simple then.

 

[hede]

I never heard of Maemo but it sounds good. Its debian and gnome base sounds even better

Wikipedia:

Despite being based on Linux and open source software in general, some parts of the Maemo remain closed source. Particularly the Wifi drivers, which contain a binary blob, and some user-space software, like certain statusbar and taskbar applets (including the display brightness applet) and applications.

Those applets will be Nokia specific? So they have to be exchanged anyway...
I do not know which wifi chip the Pandora will have, but maybe there is also no full opened source driver available. Someone stated the PowerVR will also have some binary blobs... so... this is relative!?

I think I can live with it.

 

[Tor]

The kernel and the rest can be kept separated, so it does not matter if the Nokia maemo kernel is incompatible with the Pandora wi-fi (etc) hardware. Some of the user interface applets on maemo are closed as well, but there are a couple of options: It would really be to the best if the binary distro could run directly (Debian armel, as mentioned), then applications could be downloaded and installed directly, and secondly, there are user provided open source replacements for some of those closed applets (e.g. display brightness applet)).

(The applets for controlling certain hardware aspects may have to be replaced in any case though, unless the Pandora kernel-for-maemo is set up to mimic the interfaces that Nokia use (some are in virtual filesystems like /proc/sys IIRC). In any case this would not be a showstopper at all.)

EDIT: Because maemo is a Debian-based distro, the applications are all available through online repositories, and it's just a matter or 'apt-get install' the application you want. And then all dependencies are resolved automatically, so that you don't have to go hunt for lots of support libraries to get it working. Well, when things go as they should, that is! It would in other words make available a lot of great applications, most of them user developed&provided.

 

[todd]

+1 on adapting Maemo.

I think the key here is that the Pandora is not a desktop machine with a desktop interface. Not even a laptop. PALMtop means it's great for use on the bus, standing in line, and in other situations where space and resources (e.g., power/network connections) are limited. Places where the user might not have two hands available for typing (one's holding the device) or a flat surface for mousing.

The Maemo system was designed with precisely this context in mind. It sure would be nice to find some mutual benefit with all those enthusiastic N8xx developers!

I don't know enough about Maemo to know which parts are open and which are not. But if the parts of it analagous to a window manager / panel could be adopted on top of our own kernel/fs/Xserver, that would seem ideal. It seems like it could be a good fit.

--Todd

 

[ChairmanNow]

The whole problem with correlating Maemo with the wheel is improper. Maemo doesn't roll with everything, as it is made out of rock (part closed source.. deal with our brittle one or find your own rock).

It was my understanding (and please feel free to correct me if I'm wrong) that the close source portion was mostly related to particular hardware that Nokia uses (drivers, control panels for hardware). If this is the case, Maemo would be a good and proven foundation. Again, please let me know if I'm wrong.

No, I think Pandora should have its own. Its a cool device that deserves its own stuff.

-1 for Maemo here.

I'm certain that you'd agree that there's value in having a platform that can inherit some of the existing value of another platform. Nokia's N series is very polished, attractive and functional. They obviously invested a lot of time and effort into usability testing, UI innovations, etc. Why not enjoy some of those riches?

I mean, imagine being able to have Access' Palm Emulator (that was written for Maemo and the N series) running on our Pandoras. That'd be truly amazing. And there are many other very interesting Maemo packages available that are optimized, today, for both ARM and the Pandora screen res.

Is there any other OS out there that is such an obvious drop-in replacement?

PS: Why Maemo instead of Angstrom or pdaxrom or debian? They are much easier to port, completely open and have a solid base. Why not any of those? Have you tried them enough to just deny them outright? They are all fantastic and have a beautifully large software selection. And did I mention open? I think I did right?

It depends on what you're trying to achieve. The Zaurus platform failed, after all.

The biggest danger to a "niche" device like the Pandora is that it won't have the critical mass to keep on innovating, and to stay current. This was most certainly the case the the Zaurus (which I own, and which is gathering dust). But hitching onto Maemo, you're not only inheriting the value of that platform, you're actually increasing it with new users.

How many Angstrom/pdaxrom/OpenZaurus users/devs are there today? Is there even a fraction compared to Maemo enthusiasts?

You may not know such things, or haven't used them, but Debian has an almost complete Arm processor port of great software, requiring no changes in code to work.

That's great, but also mostly irrelevant. Debian is a terrific server platform; it is a fair desktop, but not a great one (hence the need and the rise of Ubuntu). It is also most certainly not tailored for a "palmtop" platform. To do so would very much mean re-inventing the wheel - not just from a programming point of view, but from GUI development, icon development, changing the interfaces of existing apps, etc. etc. Seems like the Maemo folks already have much of this figured out.

Ah, but neither is Pandora an internet tablet, and as such is not held in the same way. Thus is the touchscreen-centric interface of Maemo, really the best fit either?

I can't imagine why you'd say that the Pandora is not an "internet tablet". The Pandora is a superset of what the Nokia hardware is. It can easily function in the same role, what with wi-fi and bluetooth, and so much more. The hardware is so much more capable, and the emphasis on gaming is just so much more fun.

I believe that there are a lot of users that would be thrilled to have the same functionality as the Nokia tablets, and to put that touch screen to good use as a primary interface. I used my fingers with my Zodiac a lot more than I used anything else, at least in terms of navigating the UI, and I bet most other Zodiac users did the same. There's nothing more immediate than jabbing an icon with your finger.

My $.02

 

[mali]

Personally owning a n800, I would love to be able to use Maemo on the Pandora, too. It is fast and and easy to control with the touchscreen, and with the extra power of the Pandora it would be extremely snappy. Some developers from the Maemo community already ordered a Pandora and may be able to help porting. The successor of the n8x0 series will have an OMAP3, too. So it will be very likely, that OS and Software will be interchangeable between the systems with only a little work.

 

[noisome]

The whole problem with correlating Maemo with the wheel is improper. Maemo doesn't roll with everything, as it is made out of rock (part closed source.. deal with our brittle one or find your own rock). You have to drill new holes to make the wheel work, which makes it brittle since the original needed the rigidity (and quite square to begin with). Therefore the wheel will crack under the pressure.

However the real wheel is GNU/Linux, which is fairly round and quite like a rubber steel-radial tire, tried and true, solid as well as quite flexible. All you need is a wheel rim/hub to attach it to whatever "thing" you like, and the hub is custom made to fit the specific thing "perfectly" and is fairly cheap to build out of whatever is available. This is homebrew stuff which doesn't require one to make-do with what is custom built. Have an angstrom base with a custom gui which fits the pandora well is much more effective in the long run, avoiding patent and copyright issues.

This also avoids stupid decisions for Maemo upgrades that the Maemo developers could be paid to change for another Nokia device or whatever else.

No, I think Pandora should have its own. Its a cool device that deserves its own stuff.

-1 for Maemo here.

Noisome

Edited for PS

PS: Why Maemo instead of Angstrom or pdaxrom or debian? They are much easier to port, completely open and have a solid base. Why not any of those? Have you tried them enough to just deny them outright? They are all fantastic and have a beautifully large software selection. And did I mention open? I think I did right?

You may not know such things, or haven't used them, but pdaxrom is beautiful on a Sharp Zaurus C3200 (better than Nokia n810 by far I think), and uses ipkg which has quite a large software selection (so does dd-wrt, openwrt)..

You may not know such things, or haven't used them, but Angstrom/OpenZaurus was on the original Sharp Zaurus linux handheld and used ipkg which has quite a large software selection (so does dd-wrt, openwrt).

You may not know such things, or haven't used them, but Debian has an almost complete Arm processor port of great software, requiring no changes in code to work.

 

[bob]

Last thing I've heard is that they still haven't found him.

 

[mali]

The whole problem with correlating Maemo with the wheel is improper. Maemo doesn't roll with everything, as it is made out of rock (part closed source.. deal with our brittle one or find your own rock). You have to drill new holes to make the wheel work, which makes it brittle since the original needed the rigidity (and quite square to begin with). Therefore the wheel will crack under the pressure.

However the real wheel is GNU/Linux, which is fairly round and quite like a rubber steel-radial tire, tried and true, solid as well as quite flexible. All you need is a wheel rim/hub to attach it to whatever "thing" you like, and the hub is custom made to fit the specific thing "perfectly" and is fairly cheap to build out of whatever is available. This is homebrew stuff which doesn't require one to make-do with what is custom built. Have an angstrom base with a custom gui which fits the pandora well is much more effective in the long run, avoiding patent and copyright issues.

This also avoids stupid decisions for Maemo upgrades that the Maemo developers could be paid to change for another Nokia device or whatever else.

No, I think Pandora should have its own. Its a cool device that deserves its own stuff.

-1 for Maemo here.

Noisome

Edited for PS

PS: Why Maemo instead of Angstrom or pdaxrom or debian? They are much easier to port, completely open and have a solid base. Why not any of those? Have you tried them enough to just deny them outright? They are all fantastic and have a beautifully large software selection. And did I mention open? I think I did right?

You may not know such things, or haven't used them, but pdaxrom is beautiful on a Sharp Zaurus C3200 (better than Nokia n810 by far I think), and uses ipkg which has quite a large software selection (so does dd-wrt, openwrt)..

You may not know such things, or haven't used them, but Angstrom/OpenZaurus was on the original Sharp Zaurus linux handheld and used ipkg which has quite a large software selection (so does dd-wrt, openwrt).

You may not know such things, or haven't used them, but Debian has an almost complete Arm processor port of great software, requiring no changes in code to work.

It's all about freedom of choice. It's not about having only one OS, or the 'best' OS. The more different approaches the better. The more possibilities there are, the more users and developers from different directions will be attracted. That's what Pandora is about, it will be the hackers dream

 

[Coder_TimT]

+1 on adapting Maemo.

I think the key here is that the Pandora is not a desktop machine with a desktop interface. ...

--Todd

Ah, but neither is Pandora an internet tablet, and as such is not held in the same way. Thus is the touchscreen-centric interface of Maemo, really the best fit either? I say we can't really tell how well it would work until we get the units in our hands and find out. But I'm hedging my best that a Maemo type interface is still not going to work as well from a usability standpoint as something like gmenu.... Once again, only my opinion and I could very well be wrong....

 

[mali]

+1 on adapting Maemo.

I think the key here is that the Pandora is not a desktop machine with a desktop interface. ...

--Todd

Ah, but neither is Pandora an internet tablet, and as such is not held in the same way. Thus is the touchscreen-centric interface of Maemo, really the best fit either? I say we can't really tell how well it would work until we get the units in our hands and find out. But I'm hedging my best that a Maemo type interface is still not going to work as well from a usability standpoint as something like gmenu.... Once again, only my opinion and I could very well be wrong....

Well, I personally would always prefer touch control over button control. It's more intuitive and direct, no multi-pressing or scrolling needed. So for me it is natural to use the screen and not the keyboard for productivity stuff, even in a clamshell design. For gaming(RTS excluded) there are the Hardware controls.

 

[Tor]

The thread got a bit confusing there.. I saw posts quoting other posts that apparently arrived much later.

Anyway: To me it looks like some of the -1'ers don't really know neither the N8x0 series nor the maemo software. The Pandora is hardware wise very similar to the N8x0 hardware: The Pandora is like a more powerful N800 paired with the N810 with-keyboard design, but with more keys (and a different letter layout). The OMAP chipset in the Pandora is the same as what, as far as anyone knows, has been planned for the next Nokia device.

Someone also mentioned Debian armel as an alternative to maemo. The maemo software _is_ a Debian-based, armel setup, it's just not _the_ Debian armel version, mostly simply due to Debian armel using a slightly newer version of glibc (the C runtime library), but they are converging fast, as maemo is catching up. Getting something to run on maemo is basically just to get the Debian sources and recompile. Then, to polish, go on with changing the gtk code slightly in order to adhere to the Hildon user interface. Then again, there are several alternative graphical interfaces, some without Hildon (there's a full KDE version for the N8x0, for example).

I for one will definitely try to get maemo running when I get my Pandora.

 

[hells_dark]

And what about canola2 ?
http://www.youtube.com/watch?v=xm5jNUnKWPI

 

[Brockway]

I like Maemo. Wouldn't be a bad simple everyday use OS. Only problem I have with my N800 is its too slow to have multiple browser windows open while downloading, and the lack of keyboard (N810 sacrificed the SDHC slots, which was a total no-go.).

 

[ChairmanNow]

And what about canola2 ?
http://www.youtube.com/watch?v=xm5jNUnKWPI

Canola2 is a perfect example of why Maemo is perfect for Pandora. Canola2 is a super-duper media player that handles just about anything you can throw at it - do a google search to see what I'm talking about - including YouTube, UPnP streaming, Last.fm, as well as pretty much every digital video and audio format around. It also has one of the best interfaces I've seen. More info here.

Again, a perfect example of not re-inventing the wheel, inheriting the goodness that's already been developed for a very similar (although less capable) platform - and already with a GUI that's optimized for the screen resolution and the touchscreen.

I'd love to have this application on the Pandora.

 

[ChairmanNow]

The thread got a bit confusing there.. I saw posts quoting other posts that apparently arrived much later.

Anyway: To me it looks like some of the -1'ers don't really know neither the N8x0 series nor the maemo software. The Pandora is hardware wise very similar to the N8x0 hardware: The Pandora is like a more powerful N800 paired with the N810 with-keyboard design, but with more keys (and a different letter layout). The OMAP chipset in the Pandora is the same as what, as far as anyone knows, has been planned for the next Nokia device.

Someone also mentioned Debian armel as an alternative to maemo. The maemo software _is_ a Debian-based, armel setup, it's just not _the_ Debian armel version, mostly simply due to Debian armel using a slightly newer version of glibc (the C runtime library), but they are converging fast, as maemo is catching up. Getting something to run on maemo is basically just to get the Debian sources and recompile. Then, to polish, go on with changing the gtk code slightly in order to adhere to the Hildon user interface. Then again, there are several alternative graphical interfaces, some without Hildon (there's a full KDE version for the N8x0, for example).

I for one will definitely try to get maemo running when I get my Pandora.

The forum went a bit weird on me for some reason, too - stuff I posted never showed up at first, and then it did.

Can you discuss the technical challenges in getting Maemo running on the Pandora, as you seem to be very knowledgeable about it?

 

[Tor]

>Can you discuss the technical challenges in getting Maemo running on the Pandora, as you seem to be very
>knowledgeable about it?

I posted a little bit about that earlier in the thread (on Friday). I know too little about the Pandora yet, with respect to what kind of software is already available - except that I know it's supposed to use an Ångstrøm distribution to start with.

In any case: maemo runs on very similar hardware to start with (omap2, and Nokia had a Beagleboard setup to demonstrate their next-gen omap3 device, so it'll become even more similar in some respects).

So, with some assumptions, we have:
- Same CPU (ARM, compatible generation)
- Presumably different wi-fi chipset
- Presumably different bluetooth chipset
- The current LCD interface of the Nokias is different from the Pandora LCD interface
The current Nokia devices have PowerVR hardware but aren't using it due to missing drivers.
- The Nokias have dedicated zoom- and fullscreen buttons, these can simply be either mapped to other keys (on an xmodmap level) or simply ignored.

I'm thinking about this on 3 levels:
1) Kernel.
2) Application level in general
3) X-Server level
Then there are some bits and pieces here and there.

It should be straight forward to put together a kernel for Pandora, as the binary blobs needed for wi-fi, bluetooth and screen should be available already. So that's 'check'. Then, if I were to start from scratch (and not even checking what Ångstrøm is using), I would just make sure that this kernel supports ARMEL as binary excecutable format.

The above should actually take care of the support for the application level in general.

Then there's the X server. This is a bit more tricky, unless there's already an X server available (and here is where my current ignorance of Pandora and Ångstrøm shows up) that can be compiled for ARMEL and the X-libs and glibc version that maemo is using.

'Bits and pieces' are small things like the status bar applets used by maemo, but they are not critical and in any case there are open source alternatives (Nokia UI bits&pieces tend to be partly closed for the moment) for some, or they're not really important.

If the above can be done it could mean that maemo software could be installed directly, straight from the repositories. Just add something like
deb http://repository.maemo.org/ diablo free non-free
to /etc/apt/sources.list, do 'apt-get update', and start doing 'apt-get install <package name >'. Add more repositories to get access to more software.

If it can't be made fully binary compatible (e.g. due to having to rely on a binary X server which forces us to use xlib and libc versions different from what maemo uses), then it should in any case be possible to make it source code compatible - the maemo distribution is 100% Debian/GTK+ compatible after all. The final polish would be to be Hildon (UI) compatible, the Hildon framework is now available in Debian so this shouldn't be a showstopper anyway. (I compile Hildon software on a completely standard Linux desktop, in addition to the maemo 'scratchbox' cross-compilation setup, as it makes for easier development.) Hildon compatibility on the Nokias is anyway mostly about getting full-screen and zoom support (full-screen: Without a border) in the matchbox window manager. One can do without. And there are alternative window managers which don't even care.

The vast majority of applications developed specifically for the 800x480 display Nokia devices are open source. There are only a few closed-source applications, and one of them is IIRC Canola, which was mentioned by other posters.. but I believe I saw something about it going open source soon. Binary compatibility would be nice for those who would want to try the Access Palm emulator (GVM) though, or the two or three other closed-source maemo applications.

EDIT: Added a little bit.