The Danger is addressing a different market - a low-cost device with flat rate data, aimed at young people, with an emphasis on email and instant messaging.
The 7650 costs about 300 UKP on contract, or $450, and has only become available in the UK in the last couple of weeks. Other European markets are similar in timing, so I doubt you have been able to buy it for 6 months. The T68i was launched a couple of months ago, and the Trium Mondo looks so horrible that a local retailer has it on special offer as obsolete stock.
None of the devices you mention are suitable for the youth market, because they are way too expensive. In time, the mainstream GPRS/MMS phones will get cheaper and address this market, but Danger has taken a clever approach by moving much of the functionality onto servers, reducing the cost of the device and enabling it to hit a low required price point.
I agree about the fashion appeal, though - the Danger looks quite drab by comparison to many new phones, so it will have to have killer features and pricing to succeed.
That's great until you go out of coverage or the wireless LAN breaks... Syncing is not going to go away, it's just going to become a continuous background activity, ensuring your data is always accessible on your device and always backed up on a server.
Even if you have great wireless LAN/WAN (802.11b/Bluetooth/3G/GPRS) connectivity, there will always be times when the wireless link is down, or simply too congested to be used (happens a lot at peak times and in busy cells with GPRS at present, since it competes with profitable voice calls).
The ideal wireless app, IMO, does background syncing, so that at any time you can (check email/check calendar/find contacts/etc). Vindigo and AvantGo for the Palm work like this (except you need to tell them to sync), meaning you always have Vindigo's database of restaurants/bars/movies (including local cinemas), and Avantgo's news/magazine pages. Very convenient...
RIM's Blackberry email device works like this, meaning you can write emails when out of coverage and it will then send/receive emails as soon as it gets into coverage. I think this is one reason it is so successful - the wireless connection is automatic and it essentially hides any coverage issues from the user.
I agree. I'm very tempted to get a Zaurus so I can do Perl development under Apache... You could even do mod_perl in your pocket - a far cry from the Palm where only mickey-mouse web servers are available, and no Perl.
I don't have a Zaurus, but Bluetooth has proved to be a good way to connect a Palm and an Ericsson T68 phone (running GPRS, a bit like CDMA2000 only slower, i.e. both are always-on packet based technologies). It's nice to be able to connect to the Internet without taking the phone out of my pocket. However, make sure you get a phone with recent firmware that can be re-flashed when necessary - my Nov 2001 vintage T68 frequently needs rebooting due to a flaky GPRS stack, and CDMA2000 stacks will probably be similar.
There are several Bluetooth stacks for Linux, and quite a few Bluetooth CF cards, so I would expect that this is feasible on the Zaurus (if not now, then soon). There is a GPRS HOWTO for Linux that can probably be adapted for CDMA2000 - if the latter works like GPRS, you run a PPP connection from the Linux system to the phone, which terminates the PPP there and then routes packets over GPRS/CDMA2000. With GPRS, you have an IP tunnel that stays up, so you keep the same IP address, albeit dynamically allocated, as you move between cells - the tunnel is re-built as you move from one local GPRS node (SGSN) to another.
CDMA2000 has a simpler architecture that may not need all this tunnelling (IIRC) and is more IP centric. It probably has a session concept, like GPRS, in that the wireless operator needs to authenticate you for billing purposes - but you just pay for bytes transferred, not for the time the session is open.
One thing to watch out for is that the PPP session, and hence the CDMA2000 session, is kept open when you turn the Zaurus off - Palm devices are stuck in the 2G world at present and disconnect the PPP & GPRS sessions for no good reason when you hit the Power button. Apparently Pocket PC devices do this better, which makes much more sense for GPRS/CDMA2000 - creating a GPRS session can take 10-15 seconds and sometimes fails, whereas sending a packet on an open session should take less than a second (takes a bit of time to acquire radio medium for the first packet in a while, subsequent packets are faster).
Hopefully some of this is useful background. Linux is probably a great platform for this sort of thing, because it is so open to tweaking and experimentation, and of course has a lot of IP applications already ported. Don't expect to run servers though - all GPRS phones are behind NAT devices due to the sheer volume of always-on users. At least until we get IPv6, which is probably only with much later 3G releases (CDMA2000, and UMTS in Europe/Asia)...
One of the better articles I've seen on how to stop spam covers Social and technical measures (Google cache), by Richard Jones - using Google because that site isn't reachable right now. It doesn't have all the answers, but has some very good ideas. Most importantly, they can be implemented by ISPs without legislation, important though that is in the medium term.
I think a combination of strong filtering, strong terms of service (e.g. take credit card numbers of those who sign up for email service, and have an automatic and substantial fine for abuse), and legislation could really help. Spammers moving offshore actually makes filtering easier, for those people who don't do a lot of business with China at any rate...
One key point is that spam-filtering should be controllable by the individual, to allow people to make sure they receive email that might look like spam (e.g. most commercial newsletters) and server-based so that nobody needs to download spam over slow dialup or mobile wireless connections. SpamAssassin is the best tool I've found so far.
The standard approach is to say that a quorum is required - say Q sites out of N, where Q > N/2. It's still quite painful to actually implement distributed and replicated databases (they were all the rage in the 80s and early 90s but never really took off), but it should be possible, particularly if a human is available to resolve occasional update conflicts.
Each site should be able to determine independently if it is part of a quorum, even in the event of network partitions (Internet breaks connectivity between two subsets of the sites). So I don't see this as a big problem, although other problems certainly exist.
There are still interesting issues with two-phase commit, where the transaction coordinator (which collects all the 'ready' responses and makes an atomic decision to proceed) is a single point of failure, but I think Transarc (taken over by IBM) may have done something in this area by moving this to a more reliable server.
Perhaps someone with more recent involvement in distributed DBs can comment.
I suppose a Google search was more difficult than speculating... This has nothing to do with Japanese - not surprisingly since iDEN was invented by Motorola. See http://idenphones.motorola.com/iden/what_is_iden.j sp for the acronym and some introductory info.
Despite the article calling it a PDA phone, it is what's called a featurephone, similar to the Ericsson T68i. If you are on iDEN it sounds like a reasonable phone, but I would want Bluetooth so I could use a wireless headset (Motorola makes a good one that's the size of four or five postage stamps).
And of course the OS is not written in Java, it is just running J2ME apps (hopefully MIDP ones). For some apps for such Java phones, see www.midlet.org and www.mobiled.net.
Yes, Win2000 is so stable on my laptop that Cygwin (a mere DLL and userland.exe's) can frequently crash it (usually with vim) since a few months ago. This should not be possible in a modern OS... Of course this is only one data point, and Win2000 is better than NT, but this is on a laptop that has up to date virus scanning and has had no new drivers installed in 18 months - so how come the OS has got into such a state?
I haven't used SQL Server for a long time, but when I did it was the only DBMS that students on a course I was teaching were able to find serious bugs in (SQL returning invalid results).
This is simply wrong - last time this came up on Slashdot, someone produced stats showing that Finland has a lower population density than the US, and of course it uses GSM quite successfully.
GSM is not going to go away! Those GSM operators who have a 3G license (about 100 of them, out of hundreds of GSM operators) will deploy overlay UMTS/W-CDMA networks, but an absolutely critical feature of these networks and any 3G phones will be seamless roaming to GSM.
GSM has over 70% of the world market, and UMTS (or CDMA2000 1x etc) will not have anything like universal coverage for a long time... CDMA is more spectrum-efficient than GSM, but GSM is going to stay around particularly in rural areas where large cells are important and 3G won't have that sort of coverage. CDMA2000 1x is an easy upgrade from cdmaOne, but going to 1xEV-DO/DV (the true 3G versions) will be a similarly expensive operation.
GSM was decreed by the European standards bodies, but it has been an incredible success - you can use GSM phones in almost every country in the world, on over 400 networks. Call quality is great, coverage is good wherever I've been (including parts of India), and you have universal services such as short message service (text messaging).
You will need new kit, but real soon now it should be possible to implement 'software radio', in which you can download new software to implement whatever new radio interface someone has dreamt up. May take a few years, but in the longer term you could just download a GSM or UMTS module to your cell phone before travelling to Europe.
GSM does not make very efficient use of spectrum - while it is very handy to be able to use my GSM phone almost everywhere in the world, most GSM operators are having to upgrade to the CDMA-based UMTS (aka W-CDMA) in order to use spectrum more efficiently.
GSM works well, but suggesting it as a solution for spectrum efficiency is quite bizarre, particularly when cdmaOne (used by Sprint PCS and Verizon in the US) is more spectrum-efficient.
The DoD no doubt wastes a vast amount of money - however that has very little to do with DARPA, which is a research agency funded by the DoD. Of course, ARPANET came from DARPA and was the essential precursor to the Internet. In any case, the DoD has good reason to use spectrum more efficiently - the less spectrum each person uses, the more people can communicate with high-bandwidth data in the same limited chunk of spectrum.
Efficient spectrum usage is probably more useful for civilian use, though - imagine thousands of people trying to meet up with friends at a football match. Typically, cell phones don't work at all well in such huge densities - efficient spectrum would help in this scenario.
Like it or not, VoIP is being used already for international calls - you may already be using it without knowing in some cases. The bandwidth is something of a red herring - they are massively expanding their IP networks using SONET/SDH rings and/or DWDM in order to carry a great deal more data traffic, so adding VoIP alongside this is relatively easy (much smaller amount of bandwidth). Latency is not a problem given suitable QoS and link-level fragmentation on slow speed links - in fact some VoIP carriers even use multiple Internet links, switching between them as required to maintain latency (though personally I'd prefer a non-Internet VoIP network, and most carriers will use this).
VoIP is also mandated by the UMTS Release 4 and 5 standards - some operators will launch with Release 4, so they will be using VoIP for cellular calls.
The SonyEricsson joint venture seems to have sorted out the phone side - the T68 and T68i are big hits in Europe, and far better than the preceding Ericsson or Sony phones (small, light, colour, GPRS, Bluetooth, and T68i has MMS for multimedia messaging). Also the recently announced phones seem pretty good.
I got a T68 in Nov 2001 and within 3-6 months about 10-15 colleagues had bought one.
802.11b does need extensions for QoS, which the IEEE is working on - this is because it uses CSMA/CA (Carrier Sense Multiple Access/Collision Avoidance) in which the transmitters sometimes need to 'back off'. A QoS scheme would let certain bandwidth be 'booked' so that there is never the need to back off as long as you stick to what you booked. The isochronous refers to a slot coming up at a fixed interval, allowing you to send at a guaranteed rate.
This is mainly important for multimedia, though perhaps also for games - one application may be wireless-linked controllers, though I'd expect ad-hoc wireless-LAN parties using consoles are more of a target. The article is a bit thin on why exactly they did this.
Re:First 32-bit processor came out in 1995?!?!?
on
AMD's 64-Bit Chip
·
· Score: 2
The first 32 bit processor came out in 1964, if not earlier - that's when IBM announced the System/360, a 32 bit mainframe whose instruction set lives on to this day in System/390 and zSeries mainframes... Backward compatibility writ large - see http://www-1.ibm.com/ibm/history/history/year_1964.html
Slashdot Mirror
The Danger is addressing a different market - a low-cost device with flat rate data, aimed at young people, with an emphasis on email and instant messaging.
The 7650 costs about 300 UKP on contract, or $450, and has only become available in the UK in the last couple of weeks. Other European markets are similar in timing, so I doubt you have been able to buy it for 6 months. The T68i was launched a couple of months ago, and the Trium Mondo looks so horrible that a local retailer has it on special offer as obsolete stock.
None of the devices you mention are suitable for the youth market, because they are way too expensive. In time, the mainstream GPRS/MMS phones will get cheaper and address this market, but Danger has taken a clever approach by moving much of the functionality onto servers, reducing the cost of the device and enabling it to hit a low required price point.
I agree about the fashion appeal, though - the Danger looks quite drab by comparison to many new phones, so it will have to have killer features and pricing to succeed.
Nextel are migrating to CDMA, because iDEN is a Motorola only standard without much of a future.
That's great until you go out of coverage or the wireless LAN breaks... Syncing is not going to go away, it's just going to become a continuous background activity, ensuring your data is always accessible on your device and always backed up on a server.
Even if you have great wireless LAN/WAN (802.11b/Bluetooth/3G/GPRS) connectivity, there will always be times when the wireless link is down, or simply too congested to be used (happens a lot at peak times and in busy cells with GPRS at present, since it competes with profitable voice calls).
The ideal wireless app, IMO, does background syncing, so that at any time you can (check email/check calendar/find contacts/etc). Vindigo and AvantGo for the Palm work like this (except you need to tell them to sync), meaning you always have Vindigo's database of restaurants/bars/movies (including local cinemas), and Avantgo's news/magazine pages. Very convenient...
RIM's Blackberry email device works like this, meaning you can write emails when out of coverage and it will then send/receive emails as soon as it gets into coverage. I think this is one reason it is so successful - the wireless connection is automatic and it essentially hides any coverage issues from the user.
I agree. I'm very tempted to get a Zaurus so I can do Perl development under Apache... You could even do mod_perl in your pocket - a far cry from the Palm where only mickey-mouse web servers are available, and no Perl.
I don't have a Zaurus, but Bluetooth has proved to be a good way to connect a Palm and an Ericsson T68 phone (running GPRS, a bit like CDMA2000 only slower, i.e. both are always-on packet based technologies). It's nice to be able to connect to the Internet without taking the phone out of my pocket. However, make sure you get a phone with recent firmware that can be re-flashed when necessary - my Nov 2001 vintage T68 frequently needs rebooting due to a flaky GPRS stack, and CDMA2000 stacks will probably be similar.
There are several Bluetooth stacks for Linux, and quite a few Bluetooth CF cards, so I would expect that this is feasible on the Zaurus (if not now, then soon). There is a GPRS HOWTO for Linux that can probably be adapted for CDMA2000 - if the latter works like GPRS, you run a PPP connection from the Linux system to the phone, which terminates the PPP there and then routes packets over GPRS/CDMA2000. With GPRS, you have an IP tunnel that stays up, so you keep the same IP address, albeit dynamically allocated, as you move between cells - the tunnel is re-built as you move from one local GPRS node (SGSN) to another.
CDMA2000 has a simpler architecture that may not need all this tunnelling (IIRC) and is more IP centric. It probably has a session concept, like GPRS, in that the wireless operator needs to authenticate you for billing purposes - but you just pay for bytes transferred, not for the time the session is open.
One thing to watch out for is that the PPP session, and hence the CDMA2000 session, is kept open when you turn the Zaurus off - Palm devices are stuck in the 2G world at present and disconnect the PPP & GPRS sessions for no good reason when you hit the Power button. Apparently Pocket PC devices do this better, which makes much more sense for GPRS/CDMA2000 - creating a GPRS session can take 10-15 seconds and sometimes fails, whereas sending a packet on an open session should take less than a second (takes a bit of time to acquire radio medium for the first packet in a while, subsequent packets are faster).
Hopefully some of this is useful background. Linux is probably a great platform for this sort of thing, because it is so open to tweaking and experimentation, and of course has a lot of IP applications already ported. Don't expect to run servers though - all GPRS phones are behind NAT devices due to the sheer volume of always-on users. At least until we get IPv6, which is probably only with much later 3G releases (CDMA2000, and UMTS in Europe/Asia)...
One of the better articles I've seen on how to stop spam covers Social and technical measures (Google cache), by Richard Jones - using Google because that site isn't reachable right now. It doesn't have all the answers, but has some very good ideas. Most importantly, they can be implemented by ISPs without legislation, important though that is in the medium term.
I think a combination of strong filtering, strong terms of service (e.g. take credit card numbers of those who sign up for email service, and have an automatic and substantial fine for abuse), and legislation could really help. Spammers moving offshore actually makes filtering easier, for those people who don't do a lot of business with China at any rate...
One key point is that spam-filtering should be controllable by the individual, to allow people to make sure they receive email that might look like spam (e.g. most commercial newsletters) and server-based so that nobody needs to download spam over slow dialup or mobile wireless connections. SpamAssassin is the best tool I've found so far.
Sounds like the Indian ISP association is wholly against this scheme, so mod the parent up...
You can comment on their proposal (and add your support if you agree with their approach) at http://not.invisible.net
The standard approach is to say that a quorum is required - say Q sites out of N, where Q > N/2. It's still quite painful to actually implement distributed and replicated databases (they were all the rage in the 80s and early 90s but never really took off), but it should be possible, particularly if a human is available to resolve occasional update conflicts.
Each site should be able to determine independently if it is part of a quorum, even in the event of network partitions (Internet breaks connectivity between two subsets of the sites). So I don't see this as a big problem, although other problems certainly exist.
There are still interesting issues with two-phase commit, where the transaction coordinator (which collects all the 'ready' responses and makes an atomic decision to proceed) is a single point of failure, but I think Transarc (taken over by IBM) may have done something in this area by moving this to a more reliable server.
Perhaps someone with more recent involvement in distributed DBs can comment.
Just wanted you to know that 'pression' is a draft beer in France - good to work some alcohol in there somehow :-)
I suppose a Google search was more difficult than speculating... This has nothing to do with Japanese - not surprisingly since iDEN was invented by Motorola. See http://idenphones.motorola.com/iden/what_is_iden.j sp for the acronym and some introductory info.
This is typically the T9 system - virtually mandatory on European GSM phones.
Despite the article calling it a PDA phone, it is what's called a featurephone, similar to the Ericsson T68i. If you are on iDEN it sounds like a reasonable phone, but I would want Bluetooth so I could use a wireless headset (Motorola makes a good one that's the size of four or five postage stamps).
And of course the OS is not written in Java, it is just running J2ME apps (hopefully MIDP ones). For some apps for such Java phones, see www.midlet.org and www.mobiled.net.
Yes, Win2000 is so stable on my laptop that Cygwin (a mere DLL and userland .exe's) can frequently crash it (usually with vim) since a few months ago. This should not be possible in a modern OS... Of course this is only one data point, and Win2000 is better than NT, but this is on a laptop that has up to date virus scanning and has had no new drivers installed in 18 months - so how come the OS has got into such a state?
I haven't used SQL Server for a long time, but when I did it was the only DBMS that students on a course I was teaching were able to find serious bugs in (SQL returning invalid results).
This is simply wrong - last time this came up on Slashdot, someone produced stats showing that Finland has a lower population density than the US, and of course it uses GSM quite successfully.
GSM is not going to go away! Those GSM operators who have a 3G license (about 100 of them, out of hundreds of GSM operators) will deploy overlay UMTS/W-CDMA networks, but an absolutely critical feature of these networks and any 3G phones will be seamless roaming to GSM.
GSM has over 70% of the world market, and UMTS (or CDMA2000 1x etc) will not have anything like universal coverage for a long time... CDMA is more spectrum-efficient than GSM, but GSM is going to stay around particularly in rural areas where large cells are important and 3G won't have that sort of coverage. CDMA2000 1x is an easy upgrade from cdmaOne, but going to 1xEV-DO/DV (the true 3G versions) will be a similarly expensive operation.
GSM was decreed by the European standards bodies, but it has been an incredible success - you can use GSM phones in almost every country in the world, on over 400 networks. Call quality is great, coverage is good wherever I've been (including parts of India), and you have universal services such as short message service (text messaging).
You will need new kit, but real soon now it should be possible to implement 'software radio', in which you can download new software to implement whatever new radio interface someone has dreamt up. May take a few years, but in the longer term you could just download a GSM or UMTS module to your cell phone before travelling to Europe.
GSM does not make very efficient use of spectrum - while it is very handy to be able to use my GSM phone almost everywhere in the world, most GSM operators are having to upgrade to the CDMA-based UMTS (aka W-CDMA) in order to use spectrum more efficiently.
GSM works well, but suggesting it as a solution for spectrum efficiency is quite bizarre, particularly when cdmaOne (used by Sprint PCS and Verizon in the US) is more spectrum-efficient.
The DoD no doubt wastes a vast amount of money - however that has very little to do with DARPA, which is a research agency funded by the DoD. Of course, ARPANET came from DARPA and was the essential precursor to the Internet. In any case, the DoD has good reason to use spectrum more efficiently - the less spectrum each person uses, the more people can communicate with high-bandwidth data in the same limited chunk of spectrum.
Efficient spectrum usage is probably more useful for civilian use, though - imagine thousands of people trying to meet up with friends at a football match. Typically, cell phones don't work at all well in such huge densities - efficient spectrum would help in this scenario.
Like it or not, VoIP is being used already for international calls - you may already be using it without knowing in some cases. The bandwidth is something of a red herring - they are massively expanding their IP networks using SONET/SDH rings and/or DWDM in order to carry a great deal more data traffic, so adding VoIP alongside this is relatively easy (much smaller amount of bandwidth). Latency is not a problem given suitable QoS and link-level fragmentation on slow speed links - in fact some VoIP carriers even use multiple Internet links, switching between them as required to maintain latency (though personally I'd prefer a non-Internet VoIP network, and most carriers will use this).
VoIP is also mandated by the UMTS Release 4 and 5 standards - some operators will launch with Release 4, so they will be using VoIP for cellular calls.
The SonyEricsson joint venture seems to have sorted out the phone side - the T68 and T68i are big hits in Europe, and far better than the preceding Ericsson or Sony phones (small, light, colour, GPRS, Bluetooth, and T68i has MMS for multimedia messaging). Also the recently announced phones seem pretty good.
I got a T68 in Nov 2001 and within 3-6 months about 10-15 colleagues had bought one.
Frequency hopping and CDMA are how you avoid this - the Motorola tech probably uses both to some extent, it certainly does the former.
802.11b does need extensions for QoS, which the IEEE is working on - this is because it uses CSMA/CA (Carrier Sense Multiple Access/Collision Avoidance) in which the transmitters sometimes need to 'back off'. A QoS scheme would let certain bandwidth be 'booked' so that there is never the need to back off as long as you stick to what you booked. The isochronous refers to a slot coming up at a fixed interval, allowing you to send at a guaranteed rate.
This is mainly important for multimedia, though perhaps also for games - one application may be wireless-linked controllers, though I'd expect ad-hoc wireless-LAN parties using consoles are more of a target. The article is a bit thin on why exactly they did this.
The first 32 bit processor came out in 1964, if not earlier - that's when IBM announced the System/360, a 32 bit mainframe whose instruction set lives on to this day in System/390 and zSeries mainframes... Backward compatibility writ large - see http://www-1.ibm.com/ibm/history/history/year_1964 .html