Originally Posted by Estel

4? c. Regular fundraising, for bigger community projects, that were not possible due to partnership with Nokia (designing decent hardware?)

Originally Posted by sulu

maybe slightly off-topic:
I think if we would actually be able do do things like that we should leave Maemo as a platform completely.
Let's face it, Maemo as a software platform is so full of compromises and flaws that it's not worth using it when designing a new mobile device from scratch. The only argument that speaks for Maemo is that it's still the best platform out there because it's the only one which is half-way free and half-way alive.
There is this wild mix of software from different Debian releases somehow patched together in order to make it work. If that wouldn't be the case projects like kernel power or CSSU wouldn't even be necessary because their features would be part of the standard distribution. Porting all these problems to a completely new device doesn't seem as a good idea to me.

If we'd ever be able to design and produce our own N900 successor the software platform should be something that already exists and that's absolutely free (as in freedom). In my (biased) opinion the best idea would be to start as a Debian sub-project like there was Debian-eee for netbooks when they were new, which eventually was integrated into the Debian distribution completely. Of course when it comes to the user interface Maemo could be a role model in many aspects but if there is a platform that could actually be used as a fork source it should be Openmoko.

Originally Posted by Android_808

sulu: there is a project to get debian on n900. Can't remember the site. It's worth checking as the creator tracks hardware support in the kernel. He logs when it was incorporated into main tree, when support is due to be merged and what components are developed outside main tree.

Originally Posted by sulu

I know. You surely mean this site:
http://wiki.debian.org/pkg-n900

The problem is that for most of the N900's hardware there are only closed drivers. Up to now there isn't even a properly working X-server under a native Debian installation. The N900 might be pretty open for a phone but if you compare it to normal PC hardware it's one of the most locked down systems that exist.

But even if one could manage to integrate all the proprietary firmwares into a standard Debian kernel one might get a running system but the result would have nothing to do with what Debian stands for (e.g. DSG). Technically the result would be pretty close to running Easy Debian on a Maemo that has no other software installed.
The reason to get a native Debian running on the N900 is not a technical one (Easy Debian does a pretty good job here) but an idealistic/politic/religious one (whatever you like to call that). And that can only be fulfilled if either the existing drivers are released under a Free license or the specs of the hardware are published so that the community could write its own Free drivers.
Therefore I think if one designs a new platform from scratch one should make sure to use hardware that can run on Free drivers completely. If that's the case we don't need an operating system anymore which is closed in many aspects itself.

Originally Posted by Android_808

sulu: I had the elektranox site thats linked to in the page you gave.

https://elektranox.org/n900/kernel/status.html

With regards to X, omap-drm is slowly coming together but doesn't support HW accel 3D

Originally Posted by mr_jrt

I agree that there is a great benefit to a Debian-rebase. I don't see the binary blobs as such a big blocker though. There is form for this - things like the Nvidia binary blobs et al and of course the non-free section. We would still of course need our own kernel package though to maintain the binary interfaces until that magical day when we have enough OSS shims and replacements to update.

For all the things I might disagree with regarding the way Nemo/Mer plans does things, the hardware adaption model seems a sensible and practical one. Finding a way to manage those within the Debian infrastructure is the only way to give continued life to Maemo 5 by ensuring it can run on more devices than just the N900. Multiple kernels and binary blobs are the unavoidable price to pay for that.

Originally Posted by sulu

@Android_808:
Thanks for providing that link again! It made me have a closer look at the project and it seems like some things have changed to the better since I last checked.

Of course from a technical point of view the non-free blobs are not a blocker, but I'm one of those Debian users who don't use the system only for technical reasons. Therefore I'd appreciate it if a Maemo replacement on the N900 could run on free drivers only. But of course I'm aware that won't happen over night.
On the other hand having an extra kernel sounds pretty ugly to me because this way the project would never become part of the regular Debian distribution. I'd favor using the Debian mainline kernel plus something like a firmware-n900 package in non-free.
Frankly I have no interest in keeping Maemo alive just for its own sake. For me it has always been just a stopgap to run (Easy) Debian on this little "subnetbook" called N900 which just by chance happened to be able to replace my regular cellphone.

@some moderator:
I guess it's time to split this subthread before it get's completely off-topic.

Originally Posted by mr_jrt

I agree it's what we have to aim for, but I feel being pragmatic is the only way we'll get there. Having a system in the (current) state the pure-Debian solution is in makes it a curiosity for developer interest only...

Having an old kernel is a big problem, yes, but unavoidable if we're going to have to maintain binary compatibility. In my eyes, strictly speaking I'm not too bothered about being part of the "mainline" Debian repos, and indeed I might wonder if that was too big a constraint. What I'd be more than happy with is a relationship akin to Ubuntu's, where we share source packages, have some packages of our own, push and pull source from Debian, and have bugs linked to those in their BTS. A tight integration of separate projects, if you will, sharing Debian's policies of pushing changes upstream (to Debian) as much as possible.

Mobile devices if nothing else will always have different sensible defaults than desktop packages!

Originally Posted by mr_jrt

Having an old kernel is a big problem, yes, but unavoidable if we're going to have to maintain binary compatibility.

Originally Posted by mr_jrt

In my eyes, strictly speaking I'm not too bothered about being part of the "mainline" Debian repos, and indeed I might wonder if that was too big a constraint.

Originally Posted by mr_jrt

What I'd be more than happy with is a relationship akin to Ubuntu's, where we share source packages, have some packages of our own, push and pull source from Debian, and have bugs linked to those in their BTS.

Originally Posted by mr_jrt

Mobile devices if nothing else will always have different sensible defaults than desktop packages!

Originally Posted by Android_808

sulu: I had the elektranox site thats linked to in the page you gave.

https://elektranox.org/n900/kernel/status.html

With regards to X, omap-drm is slowly coming together but doesn't support HW accel 3D

Originally Posted by sulu

That's the root of all problems. As long as we need an old kernel to support proprietary blobs to run our hardware we will keep falling behind Debian's state of the art. One of the main problems is the phonet firmware. How long have the nitdroid people been trying to get that working?

Originally Posted by sulu

Or have a look at the Lemote Yeeloong! This netbook was shipped with some sort of Debian Lenny but the kernel included some proprietary blobs that make a complete dist-upgrade to Squeeze impossible and Wheezy will not be an option on that device. So the lifetime of these devices will be over after the support for Squeeze will be ceased. The same will happen to a "Debian-N900" that sticks to kernel 2.6.28.
It's situations like these that make me forget my pragmatism when being stopped by proprietary software requiring people to do the same work again and again stumbling in the dark.

Originally Posted by sulu

I see that as the only realistic way to have a system that has long-term support and is up to date. Imagine you'd be forced to use Sarge today!

The difference between us and Ubuntu is that we are in a tiny niche while Ubuntu is huge, in some aspects even bigger than Debian.
They talk to each other at eye level, we can't. So we either have to integrate or we won't participate.

Originally Posted by sulu

Really? What's the difference? I see two of them but none of them qualifies to accept being tied to proprietary software or having to be separated from Debian's mainline:
1. Different interfaces: That's purely a userland topic. So there's no reason why mobile packages should conflict with desktop packages. It's as easy as installing another desktop environment.
2. Different performance: Mobile devices will always be weaker than desktop computers. But they are already strong enough to run most desktop applications. I still have a desktop computer that runs Debian fine but is outperformed by my N900.

Originally Posted by mr_jrt

I appreciate where you're coming from, but I suspect we'll have to disagree We have a great team at the moment backporting things to our current kernel. As long as we have developers fulfilling that role, then it gets us something workable now that lets us work on the higher-level functionality so it'll be ready when (and realistically, if) the open source drivers arrive. What we have to accept though is that out N900s won't live forever. One day, we will have to migrate to new devices, and we obviously will if we want to have a healthy community. Debian wouldn't be such a success if it only ever supported a single Compaq model.

Originally Posted by mr_jrt

Don't forget that Debian is so very much more than just a kernel. If Debian can support a BSD kernel (or even the Hurd!) then it can support a distinct N900-kernel, a HTC-Hero-kernel, a E7-kernel, etc.

Originally Posted by mr_jrt

That's true if you artificially limit yourself to a single device. Widen the scope to a Debian-based distro optimised for ultra-mobile devices (aka. phones, and probably tablets too), and then you start to have enough mindshare to justify sitting at the big table. There are a few existing projects that may be worth collaborating with, emdebian coming first to mind, the aforementioned pure Debian N900 work for another.

Originally Posted by mr_jrt

[*]Input devices - Being usable with a mouse vs. Having a UI that is finger friendly, but can deal with a stylus and also supporting both multi as well as single touch. Look at the number of Linux apps that are easily ported to the N900, but are completely unusable due to a) no right click, b) small widgets, c) an inability to move the pointer without clicking...

Originally Posted by mr_jrt

Power - Phone batteries are much, much, much more limited than even Laptop batteries, let alone desktop computers. Software really needs to be optimised to use as little battery as possible, which is hardly a concern for desktop software.

Originally Posted by mr_jrt

Screen Size - Mobile devices have very limited screen sizes. The resolutions are usually sufficient for a lot of old desktop software, but as they're so small, they can be impossible to click on or read. This is far more than just a different WM - few UIs are built to be scalable.

Originally Posted by sulu

Well, I have to disappoint you. I totally agree with what you say here.
btw: I intend to use my N900 "forever". I had my last mobile phone for more than 5 years and that old desktop computer I was talking about will become 14 years old in summer. I expect my N900 to become at least 5 years old too and I have two replacement N900's in my closet. And by the time all of them are gone my eyes will be to bad for those little screens anyway.

The difference between Hurd/BSD and N900 is that the former actually are something special. They need their own userlands because you can't run an existing flavour of Linux binaries on them. That's totally different for the N900. It has a Cortex A8 CPU so it runs normal armel respectively armhf binaries if you use a Linux kernel.
If you want to use a BSD kernel that would be a different story because you'd need some architecture like kfreebsdarmhf but that stil doesn't make it special for the N900. It would just be a regular armel port for FreeBSD kernels which runs on the N900 too.

Originally Posted by Android_808

With regards to debian, I think given that the system needs to be stable, we need a stable base like 6.0 rather than testing. However, given the specs of the n900 and the applications likely to be run, I would want to look down the root of recompiling for optimization and to remove unwanted features compared to the original debian packages.

with regards to packages in general, I would want to define a standard UI toolkit. with Fremantle, we started out with GTK, then incorporated QT. I have no problem with either but would prefer one to reduce memory consumption.