Letux 380: Small And New Linux (qt) Phone To Come

[dhns]

• busybox is a little spartanic

More tests have shown that we can simply copy the busybox binary (!) from the Zaurus to the Letux 380, and add symbolic links for he additional commands. Some new commands that become available: df, more, basename, dirname, head, tail, which

I have not yet done real speed tests but it "feels" faster than a Zaurus.

-- hns

[dhns]

if you simple want an appliance that is used pretty much as-shipped from the factory without too much new leading-edge stuff happening, then you can stick with legacy but tried/tested stuff.

It depends on what users need.

Question to users: What do you prefer a tried/tested device or leading-edge?

-- hns

[LinuxGadget]

Sorry dhns, i really appreciate your efforts around this little phone, but i would definitely like something with kernel 2.6.++!
Honestly, i would like a tested and stable device with a rather recent kernel and enough ram (>=128MB).  

Question to users: What do you prefer a tried/tested device or leading-edge?

-- hns

[dhns]

Sorry dhns, i really appreciate your efforts around this little phone, but i would definitely like something with kernel 2.6.++!
Honestly, i would like a tested and stable device with a rather recent kernel and enough ram (>=128MB).  

Question to users: What do you prefer a tried/tested device or leading-edge?

-- hns

Looks like squaring the circle: most recent kernel plus tested and stable

I think you have the choice to wait for the Neo Freerunner End-User release. My gut feeling is that it becomes stable enough in Summer or Fall 2008 (but that is my personal opinion).

Just for curiosity: what do you expect to be better in a 2.6 kernel than 2.4? What would be missing in 2.4?

-- hns

[koen]

Missing in 2.4:

* decent powermanagement
* decent soundsystem
* EABI
* VFP support
* sane scheduler
* dyntick
* SDIO support
* Preemption
* Highres timers
* 4 years of bugfixes

etc, etc

[HoloVector]

if you simple want an appliance that is used pretty much as-shipped from the factory without too much new leading-edge stuff happening, then you can stick with legacy but tried/tested stuff.

It depends on what users need.

Question to users: What do you prefer a tried/tested device or leading-edge?

-- hns

Well, I used to fall into the leading edge category but, these days I just don't have the time to spend there so, I have been buying and using tried and tested devices.

[speculatrix]

OK, one final reason to use kernel >= 2.6.23 ?
android! so,  just provide the base system and the software will come to you.

quoting from the elinux page, you need...
*  SoC with ARM926 or higher (e.g. ARM11) (check ARM MPCore or ARM Cortex regarding TLS issue)
** Note: ARM920T with ARMv4 instruction set will not work
* You have already a recent (~2.6.23) Linux kernel with Thumb & MMU & EABI etc support running on your target
* Soc/HW has and Linux kernel supports
* Display/frame buffer (touchscreen would be good but optional). Frame buffer has to support double buffer/page flipping.
* Keyboard
* USB (optional)
* RTC (optional?)
* Serial console
* Some storage, sufficient for ~64MB, e.g. NFS or USB stick or NAND or NOR or MMC/SDcard etc. NFS would be easiest for development
* Sufficient main memory (SDRAM) >=32MB. While 32MB seems to be enough to start, system will be really slow then. Therefore 32MB is sufficient for proof of concept, but not for a usable system.

[dhns]

Generally, I see the benefits of a 2.6.x kernel, but let me compare a "Dreamliner" vs. a "737".

While the first will have the latest technology, more distance etc., the latter is robust and mature and simply works for the masses.

And, they are used for different purposes. Would anyone expect that the 737 is upgraded to the engines of a "Dreamliner" for intercontinental flights? Probably not. The 737 is a system where all components are adjusted homogenously to each other.

Coming back to the Letux 380 device. It is more like a "737", small, useable, mature. The 2.4 kernel simply works. So, why change a running system? Who would spend money for doing that? Or do it just for fun? Sadly, the current economical situation permits less and less volunteers to work on it.

So, opening Linux devices is like squaring the circle. There are basically three different approaches:

1. retrofit Linux on a Win device: the handhelds.org (OpenEZX, Xanadux, ...) approach (needs volunteers who need years to analyse the device beyond their market availability)
2. develop an open Linux hardware from scratch: the Openmoko approach (needs a lot of funding and is commercially questionable)
3. find a manufacturer who already has a working Linux device and discuss to open it: the Letux 380 approach (we can't change anything in hardware)

As you can see none of these approaches is "optimal".

-- hns

[koen]

The 2.4 kernel simply works. So, why change a running system?

Because right now your system runs much slower and consumes much more power than with a kernel that isn't from 2002.

[dhns]

The 2.4 kernel simply works. So, why change a running system?

Because right now your system runs much slower and consumes much more power than with a kernel that isn't from 2002.

Assuming that a 2002 kernel needs more power is a pure myth. This device gets up to 240min talk time and 200h standby from a 1020mAh battery. How much could a 2.6 kernel improve on that?

And, where can you really buy a 2.6 kernel device today (i.e. April 2008) that is mature and user-ready?

I would be happy to offer a newer kernel - but nobody could tell me so far how to get that without any effort or paying a lot of money towards the manufacturer. Anyway that would make the device more expensive. It would be like asking Boeing to improve the 737 with a better engine from the Dreamliner. That takes years and millions.

It is not an objection against any 2.6 kernel - it is simply that we can get a working 2.4 kernel based device now or have someone to build a 2.6 kernel in some unknown future. But, I doubt that the improvements are that important that a 2.4 kernel is not good enough for the majority of applications.

-- hns

PS: the last official patch to the 2.4.36.2 kernel series dates from 24 Feb 2008.

[daniel3000]

Yes, that's the old discussion...
I definitely fall into the category prefering mature software over new software.
dhns put it correctly in my opinion:
Having a mature kernel eliminates the need for long periods of testing, and as we saw with the Zaurus project: Once the basic system finally is quite stable and feature-rich enough to make the device really useful, the manufacturer stops making the device.

So I really like the approach of this phone here.

OTOH, a smartphone without a QWERTY keyboard is not really useful for me.
I have one now, a T-Mobile SDA running Windows Mobile.
It replaces the zaurus now in some aspects, but it is a pain.
Text can only be entered using the phone numeric keyboard and using T9.
Text entry via touchscreen, as this new phone features, makes it a bit easier, but this would not be enough for real work for me.
I need to maintain databases, enter a lot of text (writing notes / diaries / web pages / papers...) and all this should be as compatible as possible to my Windows PC at work an my mac at home.
I admit this is not an easy profile to meet with a Linux smartphone. ;-)

But, as someone else said, I only need Linux on a phone if I can really make use of the flexibility, i.e. if this is my always-with-me mini PC. If I only use it as a phone, and carry a seperate "Mini-PC" with me, I don't need Linux on that phone.

Well, my 2 cents. As always, YMMV.

Nonetheless, a very interesting phone which could be einteresting for special projects and for some (!) private users.

In order to sell 1000 of then, though, I think you will need to find a larger company using this phone for a project.

Or maybe, with good advertising, this can be made a new standard phone for companies with field staff, which must communicate with each other a lot and sync with intranet servers etc...
Like the comany I work for.
Having different systems when sitting at the customer's plant makes it difficult for us to share information and PIMs.

If we all had one kind of flexible phone, which is able to sync with the Intranet server (Lotus Domino / Notes), and which allows to enter information easily and quickly (keyboard...) etc., then we all would be much happier.
Maybe we should discuss that offline ;-)

daniel

[speculatrix]

2.6 is a proven kernel for an embedded mass-market device - ask tomtom. in fact, you can benefit from their work, as they publish their stuff.

[dhns]

2.6 is a proven kernel for an embedded mass-market device - ask tomtom. in fact, you can benefit from their work, as they publish their stuff.

Yes, 2.6 is proven. But not for this given hardware. Not even available (who has the ported the drivers?).
No, we can't benefit from their work because we simply can't spend time/money to switch the kernel.

Let's ask in a different way: how much would you pay more if this device would have a 2.6 kernel? 1 EUR, 10 EUR, 100 EUR?
How long would you be willing to wait? 1 week, 1 month, 1 year?

-- hns

[dhns]

and all this should be as compatible as possible to my Windows PC at work an my mac at home.

It uses the Qtopia Desktop allication which should IMHO run on Windows (I have Mac only).

-- hns

[speculatrix]

Let's ask in a different way: how much would you pay more if this device would have a 2.6 kernel? 1 EUR, 10 EUR, 100 EUR?
How long would you be willing to wait? 1 week, 1 month, 1 year?

I would consider a "bounty" to get a 2.6 kernel up and running to be worth about 50 euro to me in terms of the device's longevity/usefulness, and of course would expect in return the code to be fully open and also some proper hardware details so I could fix drivers (or find someone who could), and the vendor to support their drivers properly (i.e. not just hack something together and throw it out the door).