I highly recommend checking out JanosVM, and also reading Godmar Back's KaffeOS paper for more info and background on this stuff. It's super cool.
I believe the missing needed bits are the Java class verification, of which a lot of work has been done recently on Kaffe (which JanosVM is based on). Plus the Classpath libraries have improved tremendously - this stuff could be a real contender in a fairly short period of time.
This is the "Nectar".NET virtual machine written by Tim Wilkinson's new company here in Berkeley - Hive Minded, Inc. (the name is a reference to the Borg in Star Trek).
Tim is the guy who wrote Kaffe - I used to work for him. He's no longer working on Kaffe, and I'm running the project now.
He's also the guy who wrote the first Java smart card implementation for Schlumberger (now Axalto), before Sun did their own implementation -- so he knows what he's doing.:-)
Kaffe has really picked up steam in the past year (it's been around since 1996). It's definitely not dead - it's got an active developer community and over 750,000 lines of code. It can now run a huge number of Java applications. Try it out again sometime.
Ricochet and some cell phones just interface to the serial port, so they work fine. Personally, I'm in the process of making up a cable so I can use Ricochet and my Nokia cell phone (gnokii.org) for wireless stuff. The screen on the iPaq is really quite readable in direct sunlight too, which is nice.
I don't think that anybody has gotten serious
about writing a JIT for kaffe for SuperH yet, although we're certainly open to doing it if some company wants to lend a bit of support (we're pretty busy). It would probably only take a few weeks, if that. We accept patches too. It's probably easy to get it running just as an interpreter.
Just in case anybody is equating PocketLinux with Teapot, just because they sound very similar on the surface, I'd like to point out some differences.
1) PocketLinux is GPL'd. Teapot is not. It looks quite proprietary to me.
2) PocketLinux is available for download now. Teapot is not.
3) PocketLinux runs on currently available hardware, and any hardware that runs Linux into the future. Teapot doesn't seem be available for any shipping PDAs.
4) PocketLinux has screenshots on the website. Teapot doesn't.
5) PocketLinux was released last week. Teapot seems to have been announced in 1999, but there doesn't seem to have been much action since.
We've put a lot of work into PocketLinux -- it ain't vapourware (we just didn't get much documentation up yet). Check it out.
We don't currently have a terminal emulator in our demo, but I do know that there is already GPL'd Java telnet applet available, which could theoretically be made to telnet to localhost, thus giving us the same capability on PocketLinux.
We didn't get images built using the latest version of PocketLinux for the Helio in time for LinuxWorld. We'll all be back at work next week, and we should have some posted on the site then.
We have an older version that works fine, so there's no reason why the newer version won't still work - but we'll have to take out some of the cool iPaq stuff (eg. color themes, video) which isn't terribly relevent on the Helio. Most of the applications should still be there though.
I need to do a bit more work on the kernel on the Helio too. The serial driver needs a bit of fixing, and it needs a sound driver. Also, theoretically I can get the kernel to go into a true suspend mode when it's not being used, which should enable it to sleep for as long as VT-OS does (several weeks on a set of batteries).
We weren't motivated release under the GPL because we are all that interested in dual-licensing (although if somebody wants to twist our arm, I'm sure we could do that).
Instead, we chose the GPL because we want to build a real "community" around PocketLinux, and the GPL sets the ground rules in such a way that people that want to go proprietary are on their own.
Ultimately, PDA's are very personal, and we feel that the best ones will be free to their very core. Some BSD advocates feel that the GPL is more restrictive - they may have a point, but it's not terribly material for what we want to do with PocketLinux.
Just because we're located in Berkeley doesn't mean we have to tow the line on the age-old GNU vs. BSD licensing debate.:-)
We're just tracking the handhelds.org kernel sources as far as the iPaq goes. At the moment, our demos just write to a ramdisk, so you lose the data if the battery runs out and you haven't synced.
It's looking like Linux is going to have an extreme amount of configurability in how this all gets set up -- so one thing's for certain -- it will ultimately do things the way you want it to.:-)
We're running just fine, thank you, on a 75MHz machine with 2MB of flash and 8MB of ram (the $149 VTech Helio). Too processor intensive? I don't think so.
We'll openly admit that our solution might have heavier core requirements than a conventional embedded OS written from scratch (eg. PalmOS). But we've got orders of magnitude greater functionality as well (full Linux kernel, full JVM, a set of class libraries which is shared between all the apps, etc). For a buck or two extra for a CPU with an MMU and a decent clock rate, you get the ability to code at extremely high levels of abstraction. As far as tradeoffs go, I think that one's a no-brainer (it's a pity that the current generation Palms are still using a slow Dragonball).
There'll always be language wars, and everybody has their favorite. Java, as a language, isn't that bad of a choice, because it's easy, robust, full-featured, and very modular. The downsides to it are pretty minimal, and mostly theoretical. The worst downside I can think of is just the fact that the trademark is controlled by Sun, who are trying to extract a profit out of it.
Java programs are going to be slow and bulky unless they are compiled from bytecode to a PDA's
native code by another computer before being transfered to the PDA.
We're actually doing this for the core class libraries on the Helio (which is only a 75MHz machine), in order to shorten the startup time.
Re:What about PocketLinux?
on
LinuxWorld
·
· Score: 3
Hi, I've been working for Transvirtual on the PocketLinux stuff.
PocketLinux is a PDA operating system that we've been developing, based around Kaffe and an XML-based application framework.
We're announcing it at LWCE, even though I already demoed a bit of it at OLS. Even at this early stage, it's probably the best PDA OS for Linux handhelds I've seen yet.
At the show, we've been running it on the VTech Helio, and the Compaq iPaq.
We're actually selling the Helio's for $149 at the show, but it's really for developers only at this stage.
We've got a really wicked demo built for the iPaq, with themes, handwriting recognition, a notepad, article viewer, MP3 player, and even video! We've got a few Java games running too.
Unfortunately, the demo really only came together at the last minute, so the web pages haven't been updated, and we haven't posted the sources and images yet.
Jay Carlson and I have been independently working on Helio support for the LinuxCE/LinuxVR project since around January. We merged our codebases together around May. I think I was actually the first to boot the Helio with Linux (since I wrote the serial driver). Jay was mostly working with the emulator.
People subscribe to the MAPS RBL to help them identify networks and companies that have good UCE practices.
If NSI doesn't have good UCE practices, they should be in the RBL.
With competition, if I find that I can't do business with NSI because they are in the RBL (which I subscribe to), I'll find another registrar that has UCE practices I find acceptable.
The short time I worked at Pacific HiTech, I only knew a Cliff Miller, and a Craig Oda. Who's Craig Miller?
Maybe Cliff has adopted Craig as his son?:-)
Lotsa inaccuracies in there. I think a bit too much credence was giving to the C library issue. There's really only one C library. Is everybody supposed to stick with libc5 forever?
Slashdot Mirror
I highly recommend checking out JanosVM, and also reading Godmar Back's KaffeOS paper for more info and background on this stuff. It's super cool.
I believe the missing needed bits are the Java class verification, of which a lot of work has been done recently on Kaffe (which JanosVM is based on). Plus the Classpath libraries have improved tremendously - this stuff could be a real contender in a fairly short period of time.
Cool.
This is the "Nectar" .NET virtual machine written by Tim Wilkinson's new company here in Berkeley - Hive Minded, Inc. (the name is a reference to the Borg in Star Trek).
Tim is the guy who wrote Kaffe - I used to work for him. He's no longer working on Kaffe, and I'm running the project now.
He's also the guy who wrote the first Java smart card implementation for Schlumberger (now Axalto), before Sun did their own implementation -- so he knows what he's doing. :-)
Kaffe has really picked up steam in the past year (it's been around since 1996). It's definitely not
dead - it's got an active developer community and over 750,000 lines of code. It can now run a huge number of Java applications. Try it out again sometime.
If it's free only for free OS's, then it's non-free if you go by the Debian Free Software Guidelines (as I do).
Ricochet and some cell phones just interface to the serial port, so they work fine. Personally, I'm in the process of making up a cable so I can use Ricochet and my Nokia cell phone (gnokii.org) for wireless stuff. The screen on the iPaq is really quite readable in direct sunlight too, which is nice.
I don't think that anybody has gotten serious
about writing a JIT for kaffe for SuperH yet, although we're certainly open to doing it if some company wants to lend a bit of support (we're pretty busy). It would probably only take a few weeks, if that. We accept patches too. It's probably easy to get it running just as an interpreter.
> > GPL sets the ground rules in such a way that people that want to go proprietary are on their own.
> If that is your goal, what will you do when someone uses your platform w/o releasing their code?
We've got realy nasty lawyers.
Just in case anybody is equating PocketLinux with Teapot, just because they sound very similar on the surface, I'd like to point out some differences.
1) PocketLinux is GPL'd. Teapot is not. It looks quite proprietary to me.
2) PocketLinux is available for download now. Teapot is not.
3) PocketLinux runs on currently available hardware, and any hardware that runs Linux into the future. Teapot doesn't seem be available for any shipping PDAs.
4) PocketLinux has screenshots on the website. Teapot doesn't.
5) PocketLinux was released last week. Teapot seems to have been announced in 1999, but there doesn't seem to have been much action since.
We've put a lot of work into PocketLinux -- it ain't vapourware (we just didn't get much documentation up yet). Check it out.
Has anyone else noticed that the TeaFS page
has a circle for a "Lego Component Interface"?
I wonder what that does?
:-)
We don't currently have a terminal emulator in our demo, but I do know that there is already GPL'd Java telnet applet available, which could theoretically be made to telnet to localhost, thus giving us the same capability on PocketLinux.
We didn't get images built using the latest version of PocketLinux for the Helio in time for LinuxWorld. We'll all be back at work next week, and we should have some posted on the site then.
We have an older version that works fine, so there's no reason why the newer version won't still work - but we'll have to take out some of the cool iPaq stuff (eg. color themes, video) which isn't terribly relevent on the Helio. Most of the applications should still be there though.
I need to do a bit more work on the kernel on the Helio too. The serial driver needs a bit of fixing, and it needs a sound driver. Also, theoretically I can get the kernel to go into a true suspend mode when it's not being used, which should enable it to sleep for as long as VT-OS does (several weeks on a set of batteries).
We like the GPL.
:-)
We weren't motivated release under the GPL because we are all that interested in dual-licensing (although if somebody wants to twist our arm, I'm sure we could do that).
Instead, we chose the GPL because we want to build a real "community" around PocketLinux, and the GPL sets the ground rules in such a way that people that want to go proprietary are on their own.
Ultimately, PDA's are very personal, and we feel that the best ones will be free to their very core. Some BSD advocates feel that the GPL is more restrictive - they may have a point, but it's not terribly material for what we want to do with PocketLinux.
Just because we're located in Berkeley doesn't mean we have to tow the line on the age-old GNU vs. BSD licensing debate.
We're just tracking the handhelds.org kernel sources as far as the iPaq goes. At the moment, our demos just write to a ramdisk, so you lose the data if the battery runs out and you haven't synced.
:-)
It's looking like Linux is going to have an extreme amount of configurability in how this all gets set up -- so one thing's for certain -- it will ultimately do things the way you want it to.
We'll openly admit that our solution might have heavier core requirements than a conventional embedded OS written from scratch (eg. PalmOS). But we've got orders of magnitude greater functionality as well (full Linux kernel, full JVM, a set of class libraries which is shared between all the apps, etc). For a buck or two extra for a CPU with an MMU and a decent clock rate, you get the ability to code at extremely high levels of abstraction. As far as tradeoffs go, I think that one's a no-brainer (it's a pity that the current generation Palms are still using a slow Dragonball).
There'll always be language wars, and everybody has their favorite. Java, as a language, isn't that bad of a choice, because it's easy, robust, full-featured, and very modular. The downsides to it are pretty minimal, and mostly theoretical. The worst downside I can think of is just the fact that the trademark is controlled by Sun, who are trying to extract a profit out of it.
We're actually doing this for the core class libraries on the Helio (which is only a 75MHz machine), in order to shorten the startup time.
Actually, I think we've got several thousand stuffed PocketLinux penguin toys on order, so the answer is ... probably. ;-)
http://www.transvirtual.com/pocketlinux .htm
http://www.transvirtual.com/poc ketlinux-screens.htm
Hi, I've been working for Transvirtual on the PocketLinux stuff.
PocketLinux is a PDA operating system that we've been developing, based around Kaffe and an XML-based application framework.
We're announcing it at LWCE, even though I already demoed a bit of it at OLS. Even at this early stage, it's probably the best PDA OS for Linux handhelds I've seen yet.
At the show, we've been running it on the VTech Helio, and the Compaq iPaq.
We're actually selling the Helio's for $149 at the show, but it's really for developers only at this stage.
We've got a really wicked demo built for the iPaq, with themes, handwriting recognition, a notepad, article viewer, MP3 player, and even video! We've got a few Java games running too.
Unfortunately, the demo really only came together at the last minute, so the web pages haven't been updated, and we haven't posted the sources and images yet.
I'm going to try to get on that...
http://cvs.sourceforge.net/cgi-bin/cvsweb.cgi/linu x/arch/mips/config.in?cvsroot=linu x-vr
So all this stuff works, and is available now publicly. I see that VTech seems quite impressed with it too. ;-)
Cheers,
- Jim
http://www.theq.fm/pacificat/
As long as Bill pays the gas...
Cheers,
- Jim
I've tried Fido's data service (under Windows only, unfortunately). It works fine - but you only get 9600bps - on digital only.
Cheers,
- Jim
I wrote my own accounting system with a Python front-end and a Postgresql backend. It was surprisingly easy to do.
When I get some time, I'm gonna clean it up and GPL it. Right now, it's just a bunch of loose scripts custom-tailored to my own set of accounts.
I'd like to graft an emacs front-end onto it too.
Cheers,
- Jim
NSI is way too arrogant.
People subscribe to the MAPS RBL to help them
identify networks and companies that have
good UCE practices.
If NSI doesn't have good UCE practices, they
should be in the RBL.
With competition, if I find that I can't
do business with NSI because they are in the
RBL (which I subscribe to), I'll find another
registrar that has UCE practices I find
acceptable.
Cheers,
- Jim
The short time I worked at Pacific HiTech, I only knew a Cliff Miller, and a Craig Oda. Who's Craig Miller?
:-)
Maybe Cliff has adopted Craig as his son?
Lotsa inaccuracies in there. I think a bit too much credence was giving to the C library issue. There's really only one C library. Is everybody supposed to stick with libc5 forever?
Cheers,
- Jim
Lemme rephrase that. It didn't come out right
the first time.
That Red Hat pays Havoc to work on gnome-apt!
Cheers,
- Jim