The BART system was designed and intended as a driverless system. This turned out to be a bit of an overconfidence issue since the automatic systems had several kinks, the most significant of which being the train-near sensors.
The basic problem was that BART trains were designed so as to be able to detect the presence of another train relatively near ahead on the tracks. I'm not sure what the method used was, whether it was designed to simply detect objects (radar or something similar) or whether trains produced signals that the other trains picked up. Certainly detecting large objects seems insufficient for trains designed to travel at large speeds which must also operate inside tunnels etc. In any event, during hot weather outside, the sensors would have false-positive problems, detecting trains that did not exist, and would refuse to continue. To limp past this problem, drivers were necessary to take over the role of choosing when it was no longer safe to advance. By default the trains still operated autonomously in most other ways.
Casual observation indicates a number of issues with the trains are apparently mildly driver-operated. Trains which are less than full length seem to have their stop position adjusted by the 'driver' (sometimes very ineptly). The 'driver' sometimes adjusts stop times, which is quite useful for rush hour or event-related crowds, although I sometimes wish it was pre-set so people would board more efficiently. Also it seems the 'driver' has some possibility to affect train speed, as there have definitely been cases where a change of driver just north of Union City resulted in a much changed rate of progress for the duration of the trip, although I suspect this input is optional, and infrequently used.
Certainly the BART train console is relatively elaborate, but after the manner of a point-of-sale terminal, with several print-insert buttons and no visible analog inputs of any sort. When I have watched bart train 'drivers', I have certainly seen entire station-to-station journeys made with no input at all.
This _sounds_ good on the face of it, but history says otherwise.
In the late 1800s and early 1900s, the US had far and away the largest amount of public transit, larger than any country in europe. In the 19-teens more miles of streetcar track existed in the United States than in the entire rest of the world combined. Inter-city rail was commonly used, and relatively affordable and dependable as compared to many of the nations we currently associate with rail such as Germany or France.
It's hard to identify true root causes, but certainly between the 1920s and the 1950s, american culture and spending patterns had fallen so heavily into the pattern of the automobile, that much of this was lost. Some might point to the american habit of so strongly valuing the new (cf. electricity, plumbing, etc.), while others might talk about our devaluing of the collectivist, thus valuing individual transportation. Still another point of consideration is the ugly side of capitalism, when private industry and infrustructure can sometimes poorly interact. Recent examples include Enron and the California power disaster, historically one can look to rail companies and their self destructive rail non-maintenance habits.
In any event, public transit thrived despite our lack of physical density for a good 60 years, and then died. Perhaps the point could be made that it could no longer successfully compete against private transit in our relatively non-dense environment, but even the bostonwash DC corridor has very poor transit now as compared to history and yet remains rather dense.
The problem is a good deal more complex than you suggest.
To be fair, these trips on BART are in the tens of miles, people commuting into downtown san francisco from as far as 40 or miles away on 60+mph (topspeed) trains.
The relatively comparable in-city transit system in san francisco (MUNI train) cost $1.25 a trip and feature all day passes around $5 and monthly passes at reasonable prices I cannot quote.
The BART has not offered unlimited use fares within the last several decades. There _are_ deals which include ulimited city transit and a certain amount of BART usage included. Typically BART Plus is 30 in BART ridership (about 10 trips) and uliminted MUNI.
I'm pretty sure for the compute-cluster case it depends pretty heavily on what the shared data looks like. In some cases I bet something like GFS would be nearly ideal.
a list of problems as written by the now defunct (stalled you might say) BeWine project as they existed around 2000 or so.
apparently some of the problems were insoluble, with incompatable ideas of the how to allocate virtual memory space (BeOS and windows incompatable really, not WINE's doing). Some new linux kernel features are bumping into this same problem but are all optional.. for now.
It's certainly possible that OpenBeOS could change around their idea of virtual address space allocation for a new ABI, but to support BeOS binaries, the original ABI would have to maintain these same problems. Extra work for OpenBeOS, with BeWine unforthcoming I doubt it's top priority.
Computer hardware industry legend has it that the i860 design was originally created exclusively for a video controller type application, but was then modified to act as a general purpose processor because of its apparent high performance.
I don't know if it's true, but it was told to me by embedded engineers who MIGHT be echoing correct info.
Well they had processors, which is probably what is meant. Ie. it's possible that the radiation hardened cpu on the hubble is of the same arch as a 60s mainframe of some kind. I doubt it though.
Actually the FPU was not much used in the 486 era in games because it was far too slow. Floating point operations were in the tens of cycles, something like 46 for a simple divide. Generally it was quite possible to organize your math around prefab tables and simpler variants of the math in order to get massive speedups.
Only with the coming of true 3d (eg. Quake and after) did FP actually begin to appear in earnest in games,, and this was the era of the Pentium. It partly appeared because complex 3d requires more precision than integers can provide, and partly because the Pentium had a _hugely_ improved FPU.
There have been studies done (See "The Trouble With Computers") that measure people on tasks with and without computers that demonstrate that people aren't actually faster at most mundane tasks than without ANY computer AT ALL.
This is likely true for Solaris vs Linux because the open source world has already settled on a Unix Kernel they are happy with. Linux already fits the needs of most open source people better than Solaris reasonably could be made to do.
However, there is not some big Open Source Java that fits the needs of open source people better than Sun Java. I do not mean the JVM, but the whole kaboodle including all the standard libraries. For those to be communally developed would be a win, and I see no motivation for forking them save very explicitly to derail Java. Honestly, I think Microsoft is a bit more concerned with the success of C# right now.
I find using a proxy blocker far more pleasant than using in-browser blocking. This way it works across all browsers.
Also, once I've got the rewriting proxy in place, I can do lots of other nice tricks. I have a 2 gig web cache across all browsers and http getters. I have user-agent rewriting in place for stupid sites which require me to use or not use certain browsers or http agents. I rewrite the html for some sites to eliminate useless side clutter from tables and css; this is especially useful for forums where I want to actually read the text, not have all my horizontal space occupied by advertisements.
In short, there's lots of things an in-browser blocker cannot do, but I'm highly unclear as to what a proxy blocker cannot do.
I think the point Kjella was making is that web "developers" are aware of the phenomenon, and that their sites will frequently fail when they use openwin() in their code. Thus, legitimate use of web sites opening new windows is on the wane, and the reasonability of simply disabling all pop-ups is on the rise.
This makes it much easier for the individuals (not consumers) of the world to block all pop-ups without much loss, and makes it harder for the advertisers to find wiggle-room.
The link you post says that ActiveState Python for.NET is not fast enough for more than demos. Traditional CPython is default for ActiveState Python and what most people would use.
Really? I would like more information on this since I've had POTS with my DSL for no good reason for some time. Sonic.net, my ISP, originally informed me that I could not get DSL without POTS.
If this is incorrect or has changed, please give me a hint how to change my provisioning.
It's a relatively fair comparison, when you consider that heavy rail for other purposes (such as transit lines) get at most an order of magnitude difference in efficiency, and the same goes for the cars.
But this example doesn't even scratch the surface. When you compare freight conveyed crosscountry by trains versus trucks, (both of which are very much in common usage), the tonnage efficiency of the trains grows to a full 500 times that of the trucks.
The reason that the trucks remain in use is because the entirety of their system maintenance costs are bourne by the public at large via taxes. Depending upon how you count and when you look, road maintance is between 7 and 20 percent of the entire GDP.
In my professional experience in the software industry, the term is used both ways frequently. You may be correct that filesystem was originally intended to refer to the layout only, but the term has become frequently attached (without any additional linguistic ornament) to the code which accesses and modifies said disk layout.
In short, whenever discussing such topics, be sure to disambiguate, if it matters. I also feel it's of limited value to debate which of the two meanings is valid, since they are both in common usage.
Private replay of works is not an action which is controlled by copyright. Full stop.
You're referring to public performance, which it does cover. It also covers distribution.
In the iTunes case, distribution was done with respect to the copyright and the license. If you later breach the copyright, it says you lose your rights as granted by the copyright. Sadly I'm not lawyerly enough to know what that means your legal status as to the files and private use of them is. However, you aren't either.
Suffice it to say, to be legally sure, you'd probably have to consult a lawyer, and maybe a bunch of dissenting lawyers and the agencies they represent would have to consult each other and a judge and over a lawsuit.
In practical terms, of course, I really doubt anyone will ever care about your private reproduction and media shifting, so it doesn't really matter.
If true, this is quite a reversal from my days of looking at Oracle operation on Linux, when Red Hat couldn't even get it to run stably and shipped a Hubert Mantel kernel (read SuSE kernel) on their Oracle-oriented package.
I'd suspect that SuSE still performs at least at reasonable at Oracle operation as RHEL, but I'm really just going on 5 year old trends.
I was with you on the audiophile rant, but you seem pretty uninformed about technology. Redbook audio (what you get on audio CDs) is not 128Kbps. It is 44.1 * 2 * 16 or 1411.2 Kbps, approximately, plus overhead.
That is, 44.1k samples per second, at 16 bits per sample, in stereo. 128Kbps audio is an over ten fold reduction in datasize, which is why MP3 was such a miracle (over ten times data reduction for 'similar' quality) when it passed into general usage.
When mp3 initially came into usage, it was of _obvious_ inferior quality, but for many it was good enough. mp3, and similar technologies, have advanced a great deal in their methods, and now the differences are not so extreme. If anything, 128Kbps is more acceptable now than ever before, from a quality perspective. Obviously hard drive sizes and internet speeds have affected pragmatic issues.
Slashdot Mirror
The BART system was designed and intended as a driverless system. This turned out to be a bit of an overconfidence issue since the automatic systems had several kinks, the most significant of which being the train-near sensors.
The basic problem was that BART trains were designed so as to be able to detect the presence of another train relatively near ahead on the tracks. I'm not sure what the method used was, whether it was designed to simply detect objects (radar or something similar) or whether trains produced signals that the other trains picked up. Certainly detecting large objects seems insufficient for trains designed to travel at large speeds which must also operate inside tunnels etc. In any event, during hot weather outside, the sensors would have false-positive problems, detecting trains that did not exist, and would refuse to continue. To limp past this problem, drivers were necessary to take over the role of choosing when it was no longer safe to advance. By default the trains still operated autonomously in most other ways.
Casual observation indicates a number of issues with the trains are apparently mildly driver-operated. Trains which are less than full length seem to have their stop position adjusted by the 'driver' (sometimes very ineptly). The 'driver' sometimes adjusts stop times, which is quite useful for rush hour or event-related crowds, although I sometimes wish it was pre-set so people would board more efficiently. Also it seems the 'driver' has some possibility to affect train speed, as there have definitely been cases where a change of driver just north of Union City resulted in a much changed rate of progress for the duration of the trip, although I suspect this input is optional, and infrequently used.
Certainly the BART train console is relatively elaborate, but after the manner of a point-of-sale terminal, with several print-insert buttons and no visible analog inputs of any sort. When I have watched bart train 'drivers', I have certainly seen entire station-to-station journeys made with no input at all.
This _sounds_ good on the face of it, but history says otherwise.
In the late 1800s and early 1900s, the US had far and away the largest amount of public transit, larger than any country in europe. In the 19-teens more miles of streetcar track existed in the United States than in the entire rest of the world combined. Inter-city rail was commonly used, and relatively affordable and dependable as compared to many of the nations we currently associate with rail such as Germany or France.
It's hard to identify true root causes, but certainly between the 1920s and the 1950s, american culture and spending patterns had fallen so heavily into the pattern of the automobile, that much of this was lost. Some might point to the american habit of so strongly valuing the new (cf. electricity, plumbing, etc.), while others might talk about our devaluing of the collectivist, thus valuing individual transportation. Still another point of consideration is the ugly side of capitalism, when private industry and infrustructure can sometimes poorly interact. Recent examples include Enron and the California power disaster, historically one can look to rail companies and their self destructive rail non-maintenance habits.
In any event, public transit thrived despite our lack of physical density for a good 60 years, and then died. Perhaps the point could be made that it could no longer successfully compete against private transit in our relatively non-dense environment, but even the bostonwash DC corridor has very poor transit now as compared to history and yet remains rather dense.
The problem is a good deal more complex than you suggest.
To be fair, these trips on BART are in the tens of miles, people commuting into downtown san francisco from as far as 40 or miles away on 60+mph (topspeed) trains.
The relatively comparable in-city transit system in san francisco (MUNI train) cost $1.25 a trip and feature all day passes around $5 and monthly passes at reasonable prices I cannot quote.
The BART has not offered unlimited use fares within the last several decades. There _are_ deals which include ulimited city transit and a certain amount of BART usage included. Typically BART Plus is 30 in BART ridership (about 10 trips) and uliminted MUNI.
Yeah SQLite or maybe BerkeleyDB are what you need for fast single table situations.
I expect it to be several times faster than excel at 100K entries.
I'm pretty sure for the compute-cluster case it depends pretty heavily on what the shared data looks like. In some cases I bet something like GFS would be nearly ideal.
Thank you for this day's serving of libertarian nonsense.
uhh.. nope.
jrodman@Skonnos:~ >python -c 'print 1/2.54'
0.393700787402
Wine assumes posix and to a lesser extent X. BeOS is partially posix and totally not X.
Getting wine working on BeOS would likely require some work on both wine and BeOS.
More info:
It's certainly possible that OpenBeOS could change around their idea of virtual address space allocation for a new ABI, but to support BeOS binaries, the original ABI would have to maintain these same problems. Extra work for OpenBeOS, with BeWine unforthcoming I doubt it's top priority.
Computer hardware industry legend has it that the i860 design was originally created exclusively for a video controller type application, but was then modified to act as a general purpose processor because of its apparent high performance.
I don't know if it's true, but it was told to me by embedded engineers who MIGHT be echoing correct info.
That's the 960. The 860 was used, but it was no big success.
Note that these processor designs are entirely unrelated.
Well they had processors, which is probably what is meant. Ie. it's possible that the radiation hardened cpu on the hubble is of the same arch as a 60s mainframe of some kind. I doubt it though.
Rad6k maybe?
Actually the FPU was not much used in the 486 era in games because it was far too slow. Floating point operations were in the tens of cycles, something like 46 for a simple divide. Generally it was quite possible to organize your math around prefab tables and simpler variants of the math in order to get massive speedups.
Only with the coming of true 3d (eg. Quake and after) did FP actually begin to appear in earnest in games,, and this was the era of the Pentium. It partly appeared because complex 3d requires more precision than integers can provide, and partly because the Pentium had a _hugely_ improved FPU.
It's worse than you think.
There have been studies done (See "The Trouble With Computers") that measure people on tasks with and without computers that demonstrate that people aren't actually faster at most mundane tasks than without ANY computer AT ALL.
This is likely true for Solaris vs Linux because the open source world has already settled on a Unix Kernel they are happy with. Linux already fits the needs of most open source people better than Solaris reasonably could be made to do.
However, there is not some big Open Source Java that fits the needs of open source people better than Sun Java. I do not mean the JVM, but the whole kaboodle including all the standard libraries. For those to be communally developed would be a win, and I see no motivation for forking them save very explicitly to derail Java. Honestly, I think Microsoft is a bit more concerned with the success of C# right now.
I find using a proxy blocker far more pleasant than using in-browser blocking. This way it works across all browsers.
Also, once I've got the rewriting proxy in place, I can do lots of other nice tricks. I have a 2 gig web cache across all browsers and http getters. I have user-agent rewriting in place for stupid sites which require me to use or not use certain browsers or http agents. I rewrite the html for some sites to eliminate useless side clutter from tables and css; this is especially useful for forums where I want to actually read the text, not have all my horizontal space occupied by advertisements.
In short, there's lots of things an in-browser blocker cannot do, but I'm highly unclear as to what a proxy blocker cannot do.
I think the point Kjella was making is that web "developers" are aware of the phenomenon, and that their sites will frequently fail when they use openwin() in their code. Thus, legitimate use of web sites opening new windows is on the wane, and the reasonability of simply disabling all pop-ups is on the rise.
This makes it much easier for the individuals (not consumers) of the world to block all pop-ups without much loss, and makes it harder for the advertisers to find wiggle-room.
Sorry charley.
.NET is not fast enough for more than demos. Traditional CPython is default for ActiveState Python and what most people would use.
The link you post says that ActiveState Python for
Really? I would like more information on this since I've had POTS with my DSL for no good reason for some time. Sonic.net, my ISP, originally informed me that I could not get DSL without POTS.
If this is incorrect or has changed, please give me a hint how to change my provisioning.
It's a relatively fair comparison, when you consider that heavy rail for other purposes (such as transit lines) get at most an order of magnitude difference in efficiency, and the same goes for the cars.
But this example doesn't even scratch the surface. When you compare freight conveyed crosscountry by trains versus trucks, (both of which are very much in common usage), the tonnage efficiency of the trains grows to a full 500 times that of the trucks.
The reason that the trucks remain in use is because the entirety of their system maintenance costs are bourne by the public at large via taxes. Depending upon how you count and when you look, road maintance is between 7 and 20 percent of the entire GDP.
In my professional experience in the software industry, the term is used both ways frequently. You may be correct that filesystem was originally intended to refer to the layout only, but the term has become frequently attached (without any additional linguistic ornament) to the code which accesses and modifies said disk layout.
In short, whenever discussing such topics, be sure to disambiguate, if it matters. I also feel it's of limited value to debate which of the two meanings is valid, since they are both in common usage.
Private replay of works is not an action which is controlled by copyright. Full stop.
You're referring to public performance, which it does cover. It also covers distribution.
In the iTunes case, distribution was done with respect to the copyright and the license. If you later breach the copyright, it says you lose your rights as granted by the copyright. Sadly I'm not lawyerly enough to know what that means your legal status as to the files and private use of them is. However, you aren't either.
Suffice it to say, to be legally sure, you'd probably have to consult a lawyer, and maybe a bunch of dissenting lawyers and the agencies they represent would have to consult each other and a judge and over a lawsuit.
In practical terms, of course, I really doubt anyone will ever care about your private reproduction and media shifting, so it doesn't really matter.
If true, this is quite a reversal from my days of looking at Oracle operation on Linux, when Red Hat couldn't even get it to run stably and shipped a Hubert Mantel kernel (read SuSE kernel) on their Oracle-oriented package.
I'd suspect that SuSE still performs at least at reasonable at Oracle operation as RHEL, but I'm really just going on 5 year old trends.
I was with you on the audiophile rant, but you seem pretty uninformed about technology. Redbook audio (what you get on audio CDs) is not 128Kbps. It is 44.1 * 2 * 16 or 1411.2 Kbps, approximately, plus overhead.
That is, 44.1k samples per second, at 16 bits per sample, in stereo. 128Kbps audio is an over ten fold reduction in datasize, which is why MP3 was such a miracle (over ten times data reduction for 'similar' quality) when it passed into general usage.
When mp3 initially came into usage, it was of _obvious_ inferior quality, but for many it was good enough. mp3, and similar technologies, have advanced a great deal in their methods, and now the differences are not so extreme. If anything, 128Kbps is more acceptable now than ever before, from a quality perspective. Obviously hard drive sizes and internet speeds have affected pragmatic issues.
I really enjoyed the jumping puzzles in Sly Cooper. It was a platformer to begin with and they were somewhat forgiving, but they were a LOT of fun.