Did you have a better example of air conditioning and audio touch screen interface for car ? I personally like the air conditioning interface on the Prius II. I observed that passengers found it really simple to use, so it's certainly not so wrong. I think that the audio interface have more room for improvement, especially the radio. Still simple enough for a basic use.
I agree that the consumption information can be distracting. Probably why the Prius III have it in a more simplified form in the driver indicators.
And yes, the steering wheel duplication of the common functions is a joy:-)
expect it to be universal over time. But currently in a single basic PC, aside of USB, you likely have PCI express bus, SATA bus, DVI and/or HDMI and/or DisplayPort bus, and maybe a Thunderbird (or Lightning Bolt if it really exists ?) in a new machine. So actually, even if USB 3.0 is probably universal enough to replace all of them for a regular user , the market do not offer a such option yet.
Waiting for USB 3.0 instead of SATA in raw hard disk and SSD connector. Waiting for external graphic card with USB 3.0 connector and a way to make them directly drive the screen matrix (DisplayPort ?). Waiting for a standardized internal card slot to replace the PCI express internal card. Waiting for a standardized external card slot to replace the regretted Cardbus/PCMCIA external card. Waiting for years entry level keyboards that integrate a USB hub and mouses without 2 meters long cable (wireless is not the best option everywhere and use battery).
Yes, there is many things that can be technically more nice, but someone need to have the willing and the power to change the product you find on the market.
A concept like this has be studied since 1970 in Switzerland. The subject was more warm in the early 1990, with an idea of real experimentation, but cooled down when faced the complexity of the project and his hazardous profitability. There is still some trace of it on the web:
AFAIK, the in vehicle mobile is probably offline except in case of a emergency event. Since this is a emergency call, it will start scanning to find any operator and pick one, maybe based one the signal quality or other factor. So I don't think that the in vehicle mobile is bound to any operator at all. I expect that the in vehicle mobile is SIM free.
AFAIK, due to the fact that the system use an emergency call, no SIM card is required, so the cost will probably be billed to the emergency call center. I am not certain about that, Need verification.
"8.2 In-Vehicle System Devices Subject to certain standard terms and conditions (e.g., protection for Qualcomm products as to the licensees or its customers patents), Qualcomm will not charge a royalty rate for a license for its patents essential to the 3GPP eCall in-band modem standard (3GPP TS 26.267 and TS 26.268) in subscriber devices that implement such modem standard that is higher than the royalty rate that Qualcomm charges, or may in the future charge, for a license under its applicable patents for similar devices that do not implement such modem standard. For clarity, the 3GPP eCall in-band modem standard does not include cellular modem functionality or any other functionality in a handset or device."
If there wanted to make it really free, the claim would have been far more simple and shorter. My understanding is that you have to pay, no mater what, but there is no official price.
Ok, it's mandatory now, but where is the exact specification of the system ?
I was involved in a eCall prototype device a few years ago and this was a totally crap technology. Basically this was a analog software modem on top of a GSM call, witch is a pretty stupid idea, given the fact that a SMS is a lot cheaper, reliable and faster to transmit the few data that eCall require. A few company proposed algorithms that aggressively abuse the GSM compression to pass a ridicule amount of data per second. All those algorithm are patented and require to pay licence to use them. Big business and big lose of time to only provide an expensive, slow and unreliable technology that is completely obsoleted by a 20 year old SMS. I hope that there have trow all of that crap into a trash.
Completely orthogonal state machine don't have to communicate, so you should use an other process, not an other thread.
Synchronization is a desirable feature between state machines that have to communicate. Using thread make the synchronization more difficult, less predictable, and harder to verify. Now there is situation when you wants threads to fix performance, like shorter latency or multiple core distribution. But this will invariably make the system more complex, not less.
Using threads don't solve any synchronization problem ! At the best is just move it in an other place. I have also observed that more a code is secret, more likely it tend to abuse of threads.
I don't agree: sub state machine don't have to be in separate threads.Forget the theoretical illusion that "independent states" exists: in a way on is an other, sub state machines have to communicate in some way. Most of the times the sub state machines are organized into a hierarchy to compose the main state machine. In reality you have to distribute events across the hierarchy depending of state that the OS is not award of, so threads do no help. Or some sub state machines have to change depending of the state of the others, raising synchronization issues when using threads. So many, many, many nasty and difficult to reproduce bugs cam from timing issues that affect threads coherency.
Yes monster state machine is not good, but this is exactly what Alan wanted to say: if you are unable to design a good state machine (or actually a hierarchy of sub state machines) then threads will not save you but make the whole mess even worse. After 25 years of programming, I so often fix so many broken design by killing thread and replacing them by a nice hierarchy of state machines and a lot of debug. All the synchronization problem magically disappear, the code is more easy to read, use less resources, more easy to trace, and more easy to verify on a test bench.
I also do a lot of electronics and it's so obvious when you do digital hardware that synchronous state machine is the only base that permit to construct everything reliable. Every time you lost synchronization, you get a problem that need a additional circuit to keep is safe. Thread is a way to lost synchronization. Don't use them for anything other than for performance augmentation.
The trading rules was set on a time when the technology was not able to abuse the primary intend in a such big way. The rules needs to be adapted. A proposition is to use a quantum of time, for example a few minutes, between the calculation and publication of new quotation. The transaction queue must remain secret for everyone.
I don't expect that the states will be able to impose such rules, but new stock exchange place, with more fair rules, can start up and attract companies. Old one will then look like poisoned by fast trading parasites.
Yes, this was a bit provoking. But as the same time, it's true that state machine is extremely fundamental to computers while a lot of code are not well organised around a clean state machine. It's not uncommon to see programmers that like to add a thread instead of fixing the design of the state machine.
Yes this is dangerous. Especially when someone will naively translate the "Hello World" into a language that require multiple byte UTF-8 character: the string will grow without notice.
The problem is that the management competence is not enough to be a CEO. You must have a close relationship with big players in finance that will set the goal of the amount of money there expect to extract from your business. The CEO salary, from there point of view, is just a small return on the massive profit there expect to get.
So if you don't have a plan to be very profitable, you will not be selected as CEO. If you ever get selected, this is because you promise massive profit and naturally expect a big salary for the execution of this financial service. Like in the politic, this dumb system will select the people that are the most able to lie. And because there lied at the maximum, the execution will mostly be a catastrophe. The few CEO that success (mostly because there lie so well to there employees about future profit there will never get) will get an unbelievable salary to continue to generate profit and will have big opportunity from others companies. The concentration of the power into a few, if not a single, people make the financial manipulation easy. The bad new is that this problem is not going to decrease anytime soon.
"For a hospital, it is indeed necessary to have not only consistent clocks but also accurate clocks. Otherwise records from outside ambulance services cannot be correlated with records from inside emergency rooms."
Yes.
I think that PTP (IEEE 1588-2008) is more appropriate that NTP for a hospital. PTP has some advantage over NTP in this case: 1) PTP to not require any configuration on the client side (it detect the master automatically). 2) PTP master is on the hospital and can rely on NTP or GPS redundancy for all the clients belong it, also lowering the NTP traffic on the upstream servers. 3) PTP have no discontinuity when a leap second is inserted. 4) PTP is far simpler than NTP to implement (http://ptpd.sourceforge.net/), especially in embedded systems. 5) PTP is more precise that NTP, even without special hardware. 6) PTP is already used in many measure instruments from companies that often make medical instruments too.
Actually, many think that PTP require special hardware to work (master, switch, client) because of there are many marketing around those device. But as the Open Source ptpd shows, PTP work easily on any bare PC and network with a precision of usually about 100s.
Disadvantage of PTP is that the protocol exchange is typically about one per second while with NTP the protocol exchange is typically one per several minutes. This can be a bandwidth problem for large network. Specialized switch prevent this problem, but are actually priced relatively high.
NTP have the problem of discontinuing his UTC timestamp while a leap second occur and NTP do not broadcast the actual UTC-TAI offset (historically because he broadcast UTC directly but this is now more a problem that an advantage). GPS and PTP broadcast (something very closely related to) TAI and a UTC-TAI offset, witch is the right thing to provides the precise actual time without discontinuity.
But all of them, NTP, GPS and PTP, have the problem of not broadcasting the historical leap second table, making the client of those protocols alone unable to safely compute a precise date in the past. I hope next NTP protocol will broadcast TAI, and that NTP, GPS and PTP will be able one day to broadcast the leap second table. I am certain that there is still some reserved bits somewhere in those protocol to make that working.
You are probably really right on this. The danger is only increasing as now the EU has a president and that some lobbies, mainly from the financial sector, push to give to him more power. As a Swiss citizen, I would love to see the EU as a federation with proportional representation and direct democracy. Only a dream for now...
In France, at the Futuroscope, there is a experimental projection 2D at 48fps since 1988. I enjoyed it for it brightness and flicker free movement. I remember that I was thinking that any movie theater should be like this. The realism sensation is way better that for 3D at 24fps. Can't wait to see 3D at 48fps.
Slashdot Mirror
Looking at the corruption in the banks, I can't imagine the consequence if it's about uranium instead of money.
Did you have a better example of air conditioning and audio touch screen interface for car ? I personally like the air conditioning interface on the Prius II. I observed that passengers found it really simple to use, so it's certainly not so wrong. I think that the audio interface have more room for improvement, especially the radio. Still simple enough for a basic use.
I agree that the consumption information can be distracting. Probably why the Prius III have it in a more simplified form in the driver indicators.
And yes, the steering wheel duplication of the common functions is a joy :-)
expect it to be universal over time. But currently in a single basic PC, aside of USB, you likely have PCI express bus, SATA bus, DVI and/or HDMI and/or DisplayPort bus, and maybe a Thunderbird (or Lightning Bolt if it really exists ?) in a new machine. So actually, even if USB 3.0 is probably universal enough to replace all of them for a regular user , the market do not offer a such option yet.
Waiting for USB 3.0 instead of SATA in raw hard disk and SSD connector.
Waiting for external graphic card with USB 3.0 connector and a way to make them directly drive the screen matrix (DisplayPort ?).
Waiting for a standardized internal card slot to replace the PCI express internal card.
Waiting for a standardized external card slot to replace the regretted Cardbus/PCMCIA external card.
Waiting for years entry level keyboards that integrate a USB hub and mouses without 2 meters long cable (wireless is not the best option everywhere and use battery).
Yes, there is many things that can be technically more nice, but someone need to have the willing and the power to change the product you find on the market.
A concept like this has be studied since 1970 in Switzerland. The subject was more warm in the early 1990, with an idea of real experimentation, but cooled down when faced the complexity of the project and his hazardous profitability. There is still some trace of it on the web:
http://en.wikipedia.org/wiki/Swissmetro
http://www.swissmetro.ch/en/home
AFAIK, the in vehicle mobile is probably offline except in case of a emergency event. Since this is a emergency call, it will start scanning to find any operator and pick one, maybe based one the signal quality or other factor. So I don't think that the in vehicle mobile is bound to any operator at all. I expect that the in vehicle mobile is SIM free.
AFAIK, due to the fact that the system use an emergency call, no SIM card is required, so the cost will probably be billed to the emergency call center. I am not certain about that, Need verification.
Really, a such complex claim cannot be without a lock in goal:
http://ec.europa.eu/information_society/activities/esafety/doc/ecall/pos_papers_impact_assessm/qualcomm.pdf
"8.2 In-Vehicle System Devices
Subject to certain standard terms and conditions (e.g., protection for
Qualcomm products as to the licensees or its customers patents),
Qualcomm will not charge a royalty rate for a license for its patents
essential to the 3GPP eCall in-band modem standard (3GPP TS 26.267
and TS 26.268) in subscriber devices that implement such modem
standard that is higher than the royalty rate that Qualcomm charges, or
may in the future charge, for a license under its applicable patents for
similar devices that do not implement such modem standard. For clarity,
the 3GPP eCall in-band modem standard does not include cellular
modem functionality or any other functionality in a handset or device."
If there wanted to make it really free, the claim would have been far more simple and shorter. My understanding is that you have to pay, no mater what, but there is no official price.
For reference, here is the specification I have used for the work:
http://www.etsi.org/deliver/etsi_ts/126200_126299/126267/10.00.00_60/ts_126267v100000p.pdf
http://www.etsi.org/deliver/etsi_ts/126200_126299/126268/10.00.00_60/ts_126268v100000p.pdf
http://www.etsi.org/deliver/etsi_ts/126200_126299/126268/10.00.00_60/ts_126268v100000p0.zip
The Qualcomm copyright information have been removed from the earlier revision. This do not grant that the algorithm is not protected by a patent.
While searching on the Internet, there is information that the new system is based on NG112 (VoIP based as I understand) or E100 (what that ?).
Ok, it's mandatory now, but where is the exact specification of the system ?
I was involved in a eCall prototype device a few years ago and this was a totally crap technology. Basically this was a analog software modem on top of a GSM call, witch is a pretty stupid idea, given the fact that a SMS is a lot cheaper, reliable and faster to transmit the few data that eCall require. A few company proposed algorithms that aggressively abuse the GSM compression to pass a ridicule amount of data per second. All those algorithm are patented and require to pay licence to use them. Big business and big lose of time to only provide an expensive, slow and unreliable technology that is completely obsoleted by a 20 year old SMS. I hope that there have trow all of that crap into a trash.
Completely orthogonal state machine don't have to communicate, so you should use an other process, not an other thread.
Synchronization is a desirable feature between state machines that have to communicate. Using thread make the synchronization more difficult, less predictable, and harder to verify. Now there is situation when you wants threads to fix performance, like shorter latency or multiple core distribution. But this will invariably make the system more complex, not less.
Using threads don't solve any synchronization problem ! At the best is just move it in an other place. I have also observed that more a code is secret, more likely it tend to abuse of threads.
I don't agree: sub state machine don't have to be in separate threads.Forget the theoretical illusion that "independent states" exists: in a way on is an other, sub state machines have to communicate in some way. Most of the times the sub state machines are organized into a hierarchy to compose the main state machine. In reality you have to distribute events across the hierarchy depending of state that the OS is not award of, so threads do no help. Or some sub state machines have to change depending of the state of the others, raising synchronization issues when using threads. So many, many, many nasty and difficult to reproduce bugs cam from timing issues that affect threads coherency.
Yes monster state machine is not good, but this is exactly what Alan wanted to say: if you are unable to design a good state machine (or actually a hierarchy of sub state machines) then threads will not save you but make the whole mess even worse. After 25 years of programming, I so often fix so many broken design by killing thread and replacing them by a nice hierarchy of state machines and a lot of debug. All the synchronization problem magically disappear, the code is more easy to read, use less resources, more easy to trace, and more easy to verify on a test bench.
I also do a lot of electronics and it's so obvious when you do digital hardware that synchronous state machine is the only base that permit to construct everything reliable. Every time you lost synchronization, you get a problem that need a additional circuit to keep is safe. Thread is a way to lost synchronization. Don't use them for anything other than for performance augmentation.
The trading rules was set on a time when the technology was not able to abuse the primary intend in a such big way. The rules needs to be adapted. A proposition is to use a quantum of time, for example a few minutes, between the calculation and publication of new quotation. The transaction queue must remain secret for everyone.
I don't expect that the states will be able to impose such rules, but new stock exchange place, with more fair rules, can start up and attract companies. Old one will then look like poisoned by fast trading parasites.
Yes, this was a bit provoking. But as the same time, it's true that state machine is extremely fundamental to computers while a lot of code are not well organised around a clean state machine. It's not uncommon to see programmers that like to add a thread instead of fixing the design of the state machine.
Yes, you can use VLA and disable name mangling...
Still frustrating to have to use relatively non trivial constructs to do really simple basic things.
The frustration cam from things like this, not available in C++: /* Doing something with the buffer */
int toto(int len) {
char buffer[len];
}
The "extern C" statment is to use a C library from a C++ application, not to use a C++ library from a C application !
"A Computer is a state machine. Threads are for people who can't program state machines." -- Alan Cox
http://en.wikiquote.org/wiki/Alan_Cox
Right, but using C++ to do C is actually more frustrating, especially is you code a shared library intended to be used by C applications.
Yes this is dangerous. Especially when someone will naively translate the "Hello World" into a language that require multiple byte UTF-8 character: the string will grow without notice.
The problem is that the management competence is not enough to be a CEO. You must have a close relationship with big players in finance that will set the goal of the amount of money there expect to extract from your business. The CEO salary, from there point of view, is just a small return on the massive profit there expect to get.
So if you don't have a plan to be very profitable, you will not be selected as CEO. If you ever get selected, this is because you promise massive profit and naturally expect a big salary for the execution of this financial service. Like in the politic, this dumb system will select the people that are the most able to lie. And because there lied at the maximum, the execution will mostly be a catastrophe. The few CEO that success (mostly because there lie so well to there employees about future profit there will never get) will get an unbelievable salary to continue to generate profit and will have big opportunity from others companies. The concentration of the power into a few, if not a single, people make the financial manipulation easy. The bad new is that this problem is not going to decrease anytime soon.
Ooops! ./ has lost the "micron" sign before the 's'.
[...] PTP work easily on any bare PC and network with a precision of usually about 100 __micro__ second.
Sorry,
"For a hospital, it is indeed necessary to have not only consistent clocks but also accurate clocks. Otherwise records from outside ambulance services cannot be correlated with records from inside emergency rooms."
Yes.
I think that PTP (IEEE 1588-2008) is more appropriate that NTP for a hospital. PTP has some advantage over NTP in this case:
1) PTP to not require any configuration on the client side (it detect the master automatically).
2) PTP master is on the hospital and can rely on NTP or GPS redundancy for all the clients belong it, also lowering the NTP traffic on the upstream servers.
3) PTP have no discontinuity when a leap second is inserted.
4) PTP is far simpler than NTP to implement (http://ptpd.sourceforge.net/), especially in embedded systems.
5) PTP is more precise that NTP, even without special hardware.
6) PTP is already used in many measure instruments from companies that often make medical instruments too.
Actually, many think that PTP require special hardware to work (master, switch, client) because of there are many marketing around those device. But as the Open Source ptpd shows, PTP work easily on any bare PC and network with a precision of usually about 100s.
Disadvantage of PTP is that the protocol exchange is typically about one per second while with NTP the protocol exchange is typically one per several minutes. This can be a bandwidth problem for large network. Specialized switch prevent this problem, but are actually priced relatively high.
NTP have the problem of discontinuing his UTC timestamp while a leap second occur and NTP do not broadcast the actual UTC-TAI offset (historically because he broadcast UTC directly but this is now more a problem that an advantage). GPS and PTP broadcast (something very closely related to) TAI and a UTC-TAI offset, witch is the right thing to provides the precise actual time without discontinuity.
But all of them, NTP, GPS and PTP, have the problem of not broadcasting the historical leap second table, making the client of those protocols alone unable to safely compute a precise date in the past. I hope next NTP protocol will broadcast TAI, and that NTP, GPS and PTP will be able one day to broadcast the leap second table. I am certain that there is still some reserved bits somewhere in those protocol to make that working.
You are probably really right on this. The danger is only increasing as now the EU has a president and that some lobbies, mainly from the financial sector, push to give to him more power. As a Swiss citizen, I would love to see the EU as a federation with proportional representation and direct democracy. Only a dream for now...
In France, at the Futuroscope, there is a experimental projection 2D at 48fps since 1988. I enjoyed it for it brightness and flicker free movement. I remember that I was thinking that any movie theater should be like this. The realism sensation is way better that for 3D at 24fps. Can't wait to see 3D at 48fps.
Can you describes your view ?