My Top 10 list of things to fix for the next-gen Treo/Palm platform.

I’m no David Letterman, but here’s my Top 10 list of things Palm needs to fix in order to have a successful Next Generation Palm OS platform. Many of them apply to other manufacturers in the smartphone space, so it wouldn’t be a bad idea for them to pay heed; otherwise, they might not be laughing later on. So, heeeeere we go!

10. Consistant API’s between models in a product family. Here’s where Palm has been really blowing it the last few years. Many standard features were implemented in different ways on different Treo models. That just frustrates developers, and makes the end-user experience inconsistent. Imagine being a programmer and having to write code like this:


Function FlashStatusLed
If Treo650 Then
DoItThisWay
Else If Treo680 Then
WorkAroundOsBug
DoItThatWay
Else If Treo755p Then
DoItAnotherWay
End Function


Would you want to spend time writing code for each and every model in a product family, or would you rather write code for another platform that didn’t have those issues? That’s not to say that platforms like Windows Mobile and Blackberry don’t have their own problems, but generally, stuff from the same manufacturer/product family works the same way.

9. The Bluetooth A2DP (Stereo music headphone) profile should be built-in). Come on, $50 feature phones have it, so why should Treo owners who’ve already paid big bucks for their smartphone have to fork out extra for 3rd party additions to get the same functionality?

8. A USB connector with USB host capability. We’d like to be able to use our handheld computers/smartphones with printers, flash drives, digital cameras, and USB keyboards. Generic device and printer drivers
should come installed, with the option to download specific ones Over-The-Air or at the next HotSync, based on information the system gets from the USB id’s of the peripherals. USB On-the-Go could be a good way to go; one connector that changes modes depending on what device is hooked up to it.

And pleeeeeease, use industry standard USB mini connectors and not something proprietary! I personally have gone through the Palm III series, the not-so-aptly named “Universal Connector” on my Tungsten T, the Treo 600, and now the “Athena Connector” used on Treo’s and TX’s, and had to change peripherals and cables/chargers FOUR times now. In the future, I’m not buying anything else that doesn’t use USB or Bluetooth for connectivity – PERIOD! (I might make an exception for a Linux Powered Treo that uses the existing Athena Connector, since I already have the stuff).

7. Wi-Fi and better Bluetooth. Windows Mobile and Blackberries have Wi-Fi and so does the iPhone, so Palm needs to have it too if they want to stay competitive. Wi-Fi is everywhere now, and it makes more sense to use it for streaming applications than using the Carrier’s wireless, especially in places (like Canada) where we either don’t have unlimited data plans, or they’re prohibitively expensive. Bluetooth should be upgraded to a more reliable driver stack, Bluetooth 2.0+EDR, and should include the profiles people want to use like A2DP, file transfer, BT printing and wireless input like keyboards and even mice! People use their mobile devices in different ways, so they should have the freedom to use it the way they want without having to search for other software to give them the “Out of box” experience they’re expecting. Mobile computing is becoming commonplace, and some people want a familiar “desktop feel” for working with their mobile devices. If it means including generic drivers for keyboards and mice, it’s a small price to pay to include it if it helps generate more sales and a better user experience.

6. Extensible PIM apps with a Real API! Sorry, Palm, but the DataMangler (uh DataManager) patch doesn’t count! People have Instant Messengers, e-mail, and SMS/MMS as means of communicating, as well as regular cell phone and now VOIP calling capabilities on handhelds. They don’t want to have to keep their contacts in a a half dozen different applications. The PIM apps should use an extensible database system to keep all that information in one place. Something like a light-weight SQL database that can be extended. Access has done this right by using sqlite as their PIM database engine in ALP.

The PIMS should allow for linking between the various applications. Many people want to see appointments involving or linking to a certain contact or group of people. Hierarchical tasks are also something that many people want, giving them the ability to do lightweight project management on their handhelds. Custom views are also something people want; it’s their information, let them see it the way it makes sense to them. And most important, stick to the Zen of Palm; fast, intuitive, and easy to use.

5. Better power management and battery life. Granted, my Treo 650 is two years old now, and both my batteries have seen a lot of use, but I HAVE to charge my device daily if I want to get through the day (and always have since I bought it). My wife’s Blackberry Curve goes for several days without needing a charge, and that’s with it receiving e-mail throughout the day and night. I don’t get e-mail via the carrier’s network, just the phone and SMS, and I can’t make
it past 16 hours; some days it’s flat after 10 hours. It’s gotta be better than that if you want to compete with other handheld manufacturers.

Models like the 680 and the Centro have been repeatably criticized for their lack of battery life when compared to other products. With the Linux kernel, power management should be better, especially if they implement “on-demand” CPU management. When the system is idle, like waiting for input or “sleeping” with the screen off, the system can automatically under-clock the processor, and then ramp it back up when it’s needed. For example, I wrote this post using PsMemo into the Memo’s database. Even if I’m typing fast, I’m still not taxing the CPU running at 312 MHz; it would be the exact same user experience as it was on my old IIIx running at 33 MHz. While there is all sorts of software available for the Palm platform that will under/overclock based on what applications are running, this really should be an Operating System function, and not something people have to find on their own and buy. This betters the out-of-box experience, improves performance and battery life, and when properly implemented into the OS leads to increased…

4. Stability! This has been a problem with recent handhelds, especially the 700p. Palm OS Garnet doesn’t have memory protection, so badly written applications can walk all over another app, causing a crash. Combine that with the dbcache and NVFS problems on newer units, and that has some people’s units crashing or resetting several times a day. End users won’t understand why, they’ll just say the thing is crap and move to another platform. The next generation Palm OS (from Palm Inc.) is supposed to be based on Linux, which will solve a lot of these problems. The key to success, though, will be the compatibility layer, which will allow the Garnet applications to run on the Linux kernel. Not much info has come from Palm about their layer, but a fair bit is known about Accesses’ ALP. Both companies and any manufacturers using ALP should be aware that many new sales will be based on the fact that people will want to run their favourite Palm OS applications on the new device, so the compatibility layer needs to be as stable and robust as the underlying Linux OS. Again, end-users won’t care **why** something doesn’t work; if it doesn’t work or isn’t stable and reliable, they’ll just say it’s crap and go somewhere else.

3. Listen to your customers, not just the carriers! Carriers like to “customize” devices before they go on the market. In some cases, they’ve removed functionality in order to reduce their support costs, or to force the end-user to use pricey services.

Well, heeeelllllo! Wake up and smell the coffee! It’s 2008 and Ma Bell isn’t the only player in the game. With number portability becoming the norm, people are no longer tied to a particular carrier in order to keep their number. If a carrier wants to keep customers, then they need to provide value for money. End-users tell you what is of value to them, since they’re the ones that buy the phones and pay for the contracts. Ignore them at your peril (and that goes for the carriers too!)

2. Publicly acknowledge bugs, and Fix Them! Forums, mailing lists, and instant messengers were abuzz with issues from the Treo 700p, and even the 650 back in it’s day. And people are still ticked off about the stability issues of the 700p, and the lack of an update from Palm to fix them. There’s been class action lawsuits because of issues on Palm products, and these have resulted from Palm’s poor handling of bugs and hardware problems (and Palm isn’t alone in the industry in this respect). You’d think they’d learn by now…

If you want customer loyalty, show you’re trustworthy; fess up when you screw up, and fix it to the customer’s satisfaction. Whatever it costs to fix it will save you from losing customers (and gaining lawsuits!), will likely ensure those customers are repeat customers; it may even get you new ones. Generally, a happy customer tells 2 or 3 other people while an unhappy one tells on average 8-10 other people. These days with the Internet and all it’s possible soapboxes, one unhappy customer can tell large part of the world population!

1. Marketing and Advertising! It doesn’t matter if you have a superior product to the competition if nobody knows about it or has enough reasons to buy it! (Hmmm, remember Betamax?) There are whole magazines touting Windows Mobile products, MS Mobile TV commercials, Apple iPhone commercials and fanboy advertising, Blackberry TV and print ads, but where’s Palm? I haven’t seen any advertising for Palm in Canada in recent years; I can’t speak to other markets. Palm needs to get out there and advertise. Word of mouth helps lot too, so fix the other things on this top 10 list, and your happy customers will be your most successful sales force.

In the mean time, start targeting your advertising to your different market segments, and create hardware/software bundles that meet their needs. Don’t do it by staying in house; get out there and get some new blood, take some risks and be creative. Use focus groups of real, everyday people if you have too! Apple and the iPhone have set the new bar for creating a buzz for a product; either beat it, or go home.

With Palm’s stock price heading downward, and competition heating up, Palm Inc. can’t afford to go at this as they have in the past. To paraphrase Master Yoda, “Do or do not, there is no status quo”.

Originally posted to PalmAddicts here.

Advertisement

Is there a virtual Palm in your Future?

First there was virtual pets, now there is virtual Palms. With the recent release of the Garnet Virtual Machine for the Nokia Internet tablets, it now joins StyleTap as another contender for virtualizing your Palm device on different portable hardware. The real benefit of the Palm OS platform has always been the abundance of third-party applications which can personalize your handheld for your lifestyle. Many people get hooked on a particular application, which may not have a counterpart on another platform, and that confines them to staying on the same platform. Virtualization technologies like StyleTap and the Garnet VM allow you to run your favourite applications on a different hardware/OS platform, which means you can choose new hardware which has the features you want, while still being able to run your favourite applications. For those people who tire of dealing with the limitations of the existing Palm OS or hardware, you now have the freedom to choose something more to your liking.

As for the timing of this release, it’s definately a shot across Palm’s bow, and will hurt Palm’s handheld sales to some degree this holiday season. Nokia has recently released it’s N810 Internet Tablet, and the N800’s are still selling strong, priced around $260 CDN, compared to the Palm Tungsten TX priced around $280 CDN. So, for $20 less, you get 2 memory slots, a webcam, stereo speakers and headphones, and a 800×480 screen, along with Wi-Fi and Bluetooth and a Linux-based OS. Up till now, the Internet tablet was really only lacking a viable PIM suite; with the release of the Garnet VM, you can now have all your old favourites for free, and HotSync too! It also benefits Access in promoting it’s ALP product to potential clients, and by building in feedback tools to the Garnet VM, they can benefit from the “testing” by a large userbase.

So, Palm, I’ll say it again; it’s time to swallow your pride and look long and hard at licensing ALP, since your home-grown OS won’t be out ’till late 2008, or the next virtual Palm we see may be a virtual Palm Inc., as seen in the Wayback Machine.

Originally posted to Palm Addicts here.

Recent deal does not bode well for Palm

Given that Palm recently stated in the Q1 FY08 investor conference call that the next generation Linux-based Palm OS would not be ready until the end of 2008, and that it was being developed in-house, I have grave concerns that Palm will not get it out the gate in time to make it worthwhile. These concerns stem from the recent deal between Emblaze Mobile, Sharp, and ACCESS, where they announce they are developing a device which will “revolutionize mobile communication”. Emblaze claims they’ve been working on the device design for the past five years, and that Sharp will provide the hardware and ACCESS the software. Let’s look at this more closely to see why I’m concerned for Palm.

First of all, Sharp is involved. Sharp is a major manufacturer, one that makes a lot of parts for mobile devices; in fact, many Palm devices use Sharp LCD’s . They’ve also made the Zaurus line of Linux-powered PDA’s, which have quite the following in Linux circles. They’ve been making other PDA’s for years; my first PDA was the Sharp SE-300. They also manufacture Notebook PC’s, mobile digital audio players, projectors, professional video products, and other entertainment devices which are of interest to today’s consumers. So, basically they are a multimedia, computing, and handheld powerhouse, one that makes much of their own hardware, which lowers their parts cost (and thus overhead) greatly. Compared to Palm, which contracts out the building to companies like HTC, they are now at an advantage by being able to sell the devices cheaper, while still making a good profit.

Sharp is also the manufacturer of the Sidekick/Hiptop series of mobile phones, which gives them a lot of experience in the mobile communications space. Now, when you combine that experience with their other product lines, it makes them a very formidable player in the mobile space. What has been holding them back is the limited selection of 3rd party software for the Danger OS running the devices, and the fact that software is written and encrypted for specific versions of the OS on specific carriers. Not all carriers carry their devices, since they compete somewhat feature-wise with the Blackberry, which limits their brand recognition and market penetration of the mobile space.

Now, we all know that ACCESS owns the former PalmSource, and it’s present and future versions of Palm OS (Garnet and ALP). ALP, the Linux-based successor for Palm OS, has been in development for years, and is now available to licencees. The Access Developer network is open to developers, and the software development kits and compatibility test tools are available now. That means the ALP software is available to manufacturers NOW, although we just haven’t seen any devices based on it yet. So, where as Palm is still working on “their version” of Palm OS Linux, Access has ALP ready to roll.

Now, enter Emblaze Mobile, centre stage. They’re the makers of the Emblaze Touch 7, a multimedia feature phone which was targeted at the youth market in the UK and Europe. They’ve got hardware experience and software experience, much like Sharp, but they’re a small, albiet creative, player in the mobile space. Think Handspring when they first came out with the Treo 180.

When you look at them individually, it’s just “business as usual”; different players in the hi-tech space. What should have Palm very worried is this deal, making them a direct threat to the Palm Inc. device lines. Emblaze mobile is kinda like what Handspring was before Palm bought them up, bringing the Palm founders back into the fold. A small, passionate company looking to create a wave in the communications world. Sharp brings manufacturing muscle to the team, as well as their own experiences with the Sidekick/Hiptop and Zaurus lines. Access brings instant Palm OS compatibility to the mix, allowing Palm device owners an easy migration path to the Emblaze/Sharp platform, while bringing thousands of existing Palm OS applications over to their platform. This mashup platform, (part Zaurus, part cell phone, part Palm device, part multimedia platform), should have Palm very worried. All the Palm OS goodness running on a Linux Kernel, with all the hardware people have been asking for, and by the looks of it, earlier than Palm can deliver their OS. Very concerning for Palm indeed…

Now, this is all my analysis and conjecture. No specs have been released, no dates finalized. But given the fact that a) ALP has been finished for a while, b) Emblaze has been working on a hardware design for 5 years, and c) Sharp is involved, it’s really looking like a better Palm device is about to be released, but not by Palm. Of the 6 Palm OS devices I’ve owned, 3 were not designed by Palm (TRG Pro, Handera 330, and Treo 600). Many people loved the Sony Clie’s for the same reason. Those companies were making better Palm OS devices than Palm was. It’s looking like we’re going to have the same situation again in Sharp/Emblaze.

Now, what really concerns me about Palm’s decision to write their own OS instead of licensing ALP like I suggested in postponing the Foleo, is the fragmentation of the developer community. Some of the smaller developers are not certain they’re going to write for the next generation Palm OS, because they’re already tired of fixing Palm’s device-specific “improvements”. If they take the same approach with their OS, and developers are faced with a choice of developing for ALP and it’s licencees or Palm, they may choose to develop for ALP. If Sharp brings this ALP powered device to market, and it’s a better “Palm” than Palm’s product/OS, then developers will choose to go with the greater opportunites for sales. And if Sharp/Emblaze does come out hardware-wise with a “Better PalmPhone” than the Treo, the ALP Palm OS compatibility will ensure they have a best-seller on their hands. Everyone who has been frustrated with the lack of Wi-Fi, OS resets, and NVFS issues will take a good look at a product which has those issues addressed, and yet will still run their existing applications.

Palm, learn from your Cobalt mistakes; you can have a great OS concept, but if no one wants to develop for it, it’s worse than worthless – it’s a moneypit. If you’re busy writing an “ALP clone”, then it better be compatible with existing applications, and it should be compatible with applications written for Linux/ALP with a minimal amount of rewriting or just a re-compile against the different headers & includes. If you make it incompatible, you may find your products going the way of Cobalt; into the trash can. This Emblaze/Sharp/Access announcement means the pressure’s on, Palm; get it right, or don’t bother.

Originally posted to Palm Addicts here.

Advice sought from (Linux) gamers on controllers…

I know a fair number of people read this blog every week, many looking for answers to their problems, or looking for new ideas and solutions. I’m hoping some of you readers may help me by way of your own experiences.

Earlier this week, I wrote about installing Childsplay on Freevo, our Home Theatre PC. Childsplay is normally run on a PC, so it normally expects a mouse and keyboard. I’m hoping to get a game controller to be able to have my kids play with it, but I’m not really a gamer. I’m not really sure which type of controller to get, but I do know I want to get a wireless controller.

I know there are Linux drivers available for Xbox 360 Wireless controllers (xpad), the Wiimote (Cwiid), and the Playstation 3 controller. The xpad drivers have been around for a while and are pretty stable, but I don’t know as much about the Cwiid (Wiimote) or PS3 drivers. I haven’t really spent any “hands-on” time with any of the controllers, so I’d be interested in hearing people’s advice or recommendations on the controllers in general, and under Linux in particular.

I think the Wiimote concept (pointing device with optional accessories) may be the best for my application, but I’m not sure if the drivers are mature enough for use. Here are my specific questions:

How would the three of them compare as general purpose game controllers? (either under Linux or on their native platforms) I know it’s an “apples vs oranges” comparison in some ways, but I’m thinking along the lines of general compatibility/usability with a variety of educational games by non-gamer type adults and young kids.

How are the linux drivers? I’ve already compiled the ones for the Xbox 360 Wireless, so it’s straight forward. I don’t have any experience with the Cwiid or PS3 ones… Is there a mouse emulation mode for the Xbox (xpad) or PS3 drivers?

Anything I’m missing? The last time I played games, it was on a Commodore 64 or Atari 2600 (I know, I’m dating myself!)

Thanks for any and all help rendered! Please feel free to use the comments for your replies, so that others may also benefit from your advice. Much appreciated!

Have a great weekend!
/Michael

It’s Childsplay, on Freevo

Freevo has become a staple in our family’s lives, and a project I’ve enjoyed setting up for us. I have Freevo’s recordserver and webserver components running on the machine in the office, where it records the TV shows my wife and I are interested in seeing (usually after the kids go to bed). Freevo is also running on a dedicated Home Theatre PC in our family room, where it is often used to watch kid’s videos and browse family pictures and videos stored on the fileserver. It allows us to enjoy family-oriented media together as a family. My oldest child has a Leappad educational toy, but she’s getting to the age where she’s interested in more “dynamic” stuff. Now, I know there are all kinds of educational “game consoles” out there, but why would I want to hook up another gadget to the TV, especially one that requires a lot of extra software titles to make in interesting? That could get quite expensive, quite fast…

Here’s a solution that’s childs play… make use of the existing Linux-powered home theatre PC, and make it an educational console! Childsplay is a suite of educational games written in Python (the same programming language as Freevo is written in). Childsplay is available in Ubuntu’s Universe repositories (and in Debian Unstable), as well as by downloads from the sourceforge website. I enabled the games section of Freevo, and configured Childsplay as per the Linux Games Config section of the Freevo Wiki. My specific configuration was more like this…

apt-get install childsplay childsplay-plugins

In Freevo’s local_conf.py, I enabled the games plugin, and configured the games items as follows:
GAMES_ITEMS = [ ('Childsplay', '/etc/freevo/games/childsplay',
('GENERIC', '/usr/games/childsplay', '', '', [ 'childsplay' ] )) ]


And in the /etc/freevo directory, as root (or add sudo in front of each of these commands, where the # symbol is)
# mkdir /etc/freevo/games
# mkdir /etc/freevo/games/childsplay
# > /etc/freevo/games/childsplay/childsplay.childsplay


You’ll now have a menu item, “Play a game”, and if it’s the only game configured, it’ll run Childsplay. You can then pick which particular childsplay game you are interested in playing by clicking on the thumnail picture. We tried it with the wireless keyboard, and it works well. I may experiment with it’s “kiosk-mode” when I have some free time, which uses an on-screen keyboard, and experiment to see if it works with a joystick or gamepad.

Ubuntu “Fiesty Fawn” 7.04 quickies

A couple of quick notes about “Fiesty” after I upgraded the HTPC in the family room from 6.10 to 7.04:

It includes Linux kernel 2.6.20 (Debian Etch only includes 2.6.18). It’s got nice hardware support, including the k8temp module needed by my AMD 3500+ processor on the ASUS M2NPV-VM motherboard. It also has the Nvidia drivers available in the restricted modules section. I had to manually install the nvidia drivers, as it was an ‘apt-get’ upgrade (I’ve been stripping this install back to the bones since it’s more of a multi-media “appliance” than a PC).

Mplayer is majorly broken in Fiesty. Playing a video results in distorted sound. I ended up having to reconfigure Freevo to use xine instead of mplayer. So, be wary if you use mplayer (or software that depends on mplayer) if you plan on upgrading to Fiesty Fawn.

UPDATE: July 2007. The specific problem is that mplayer’s mp3lib is broken. After googling around, I found that adding “afm=libmad” (without quotes) in /etc/mplayer/mplayer.conf fixed the problem.

Freevo and IVTV 0.8 gotcha’s

I’ve been using Freevo as a “VCR” for about a year now, and I really like it. Freevo is a lightweight multi-media PC framework written in python, running on Linux, using many existing open-source multi-media players like mplayer and xine. In my case, I’ve been using it for the last year running without the GUI on my server in the office, while displaying the recorded programs using a Hauppage MediaMVP running the open-source MVPMC (MVP media center) instead of Hauppage’s Windows-based software. The MVP is great because it’s quiet and low-power, boots quickly, and produces a great video image; we use it to watch recorded programs in our bedroom after the kids are in bed. The MVPMC software really gives it a lot of flexibility (more so than Hauppage’s own software), and it’s under steady development to give it even more features. MVPMC also integrates nicely with that other Linux HTPC package (MythTV), but I personally like Freevo’s design much better. I’ve been using Hauppage’s WinTV PVR 150 as the capture card (I use the less expensive 150 OEM MCE edition, as they don’t come with the Windows software, just MCE drivers). The full retail version comes with a remote, which is supported under Linux using the LIRC package.

Main menu	TV Guide	Movie Browsing

Music Browsing	Playing Music	Image Browsing

So, getting on with the story… I noticed that Georg had packaged 1.7.0 for Debian and Ubuntu, so I decided it was time to upgrade from 1.6.3, as 1.7 has some really nice new features. I hit a couple of minor packaging issues, which I figured out and sent Georg an e-mail last night (he had them fixed in the morning! Gotta love Open-Source developers!) I backed up my local_conf.py file, and copied the new local_conf.py.example to local_conf.py, and started restoring my settings from the backup file. Generally, everything worked well, except for some video issues.

This is how Freevo’s IVTV options were configured “out-of-the-box” in the Debian packages I used (IVTV is the open-source driver effort for cards like the WinTV PVR series).

TV_IVTV_OPTIONS = {
'input' : 4,
'resolution' : '720x480',
'aspect' : 3,
'audio_bitmask' : 233,
'bframes' : 3,
'bitrate_mode' : 1,
'bitrate' : 4000000,
'bitrate_peak' : 4000000,
'dnr_mode' : 0,
'dnr_spatial' : 0,
'dnr_temporal' : 0,
'dnr_type' : 0,
'framerate' : 0,
'framespergop' : 15,
'gop_closure' : 1,
'pulldown' : 0,
'stream_type' : 10,
}

Now, in IVTV 0.8, many of the IVTV options are now configured through the command v4l2-ctl, and not ivtvctl as they were in previous versions of the IVTV utilities. As the drivers progress in developement, ivtvctl will go away, as all it’s features get rolled into the v4l (video 4 linux) drivers and utilities

Here are the defaults for the two settings that got me when I upgraded to the 0.8 drivers from the 0.7 ones.

video_aspect (menu) : min=0 max=3 default=1 value=1
video_b_frames (int) : min=0 max=33 step=1 default=2 value=2 flags=update

I needed to set ‘b frames’ to 2 to eliminate some picture jitter I was getting, and I needed to set aspect to 2 to get the aspect ratio back to the expected 4:3 for my TV (3 is 16:9 widescreen, which I don’t have). As the drivers develop and get merged into the v4l project, expect to tweak settings as defaults may change.

Here are the possible aspect ratio choices:

aspect aspect ratio
1 1:1
2 4:3 (normal TV)
3 16:9 (Widescreen)
4 2.21:1

v4l2-ctl -l | more is a helpful troubleshooting tool that lists all possible controls and their values. It lists what the defaults are as well as what the current value is, so if you run into problems, try setting your values to defaults, then changing one at a time to see if that fixes your problems (or improves the quality, etc.).

I’ll write something more about Freevo in the future (It’s a project I really enjoy using!), but if you have questions or comments, feel free to use the comments on this post to ask away!

Using a Linux server as a Bluetooth Network access point.

One astute blog reader noticed I use my Linux server for Internet access, and wrote to ask me how I did it. I’d set it up a couple of years ago to work with my Tungsten T, and I only had a minor hiccup when I tried to switch to the Treo (I can’t remember if I had the same problem with the Tungsten T, but I think so); more on that later.

The definitive guide to HotSyncing over Bluetooth can be found at pilot-link.org, in the Bluetooth HotSync How-to. Internet access is included in it under “surfing the web”. It’s a bit dated, as pilot-link and Bluez are usually included in most major distributions these days; just use your package management system to install it, instead of from source. You have to make sure ip forwarding is turned on for your kernel. On Debian/Ubuntu/Knoppix distributions, you do that by editing the /etc/network file and adding ip_forward=yes (then save). It will take effect as of your next reboot. For the impatient who want to try it without rebooting, use the following as root (or with sudo) from the commandline echo 1 > /proc/sys/net/ipv4/ip_forward

The hiccup I had with the Treo involved pairing it to the Bluetooth dongle on the server. In the network preferences for the BTLan connection I made, I had to pair it using “Connect to PC”, and then after it was paired, I had to change the setting to “Connect to LAN”. It wouldn’t pair if it was set to LAN the first time around; it may be because the Dongle isn’t advertising network access as an available service. I haven’t bothered to look into it further, as it works fine once paired as a PC.

As a SOHO network access point for a single user, it works pretty well. According to the speed reports at text.dslreports.com, the Treo’s Blazer web browser averages between 125-150 kbps. I can pick up my e-mail just fine, and local access to my fileserver over ssh (secure shell) runs great as well. More data intensive stuff like file transfers or VNC (kind of like remote desktop) are about the same speed, which is slower than Wi-Fi or a wired network, but still very usable. The trade off is speed for convenience and no cellular data fees; with Bluetooth dongles available in the $25 range, it does make for inexpensive wireless access at home.