SQL may not appear on the front end, but that may be what the server gets from the tool. Most OLAP queries aim at aggregating all the data within particular ranges, eg "all sales of books costing over $20, in stores in the Southwestern US, to customers who paid by credit card." An OLAP tool would build the query by presenting the user with choices about each dimension, and then accumulating SQL predicates to define each range, eg "price > 20", "payment_type = 'credit card'", etc. Once this process is over, the tool comes up with a whopper of a query, possibly involving dozens of predicates. The database then has to look at this beast, and without any human assistance, come up with a good way of fetching the data.
Multidimensional refers more to the physical data layout than the actual logical model, which can still be relational.
In a conventional RDBMS, a tuplet of values (a,b,c,d), say, can end up in any free block in the table (and indexes are needed for each column searched). In a multidimensional db, tuplets are laid out as if their values were array subscripts, the way a multidimensional array would be in, say, FORTRAN. It's a good deal easier to find data that way, since it's the same as an array lookup. The downside is that the data has to be fairly dense, or else you'll end up with a lot of empty holes and wasted space.
In a data warehouse, the array is usually organized along dimensions that are well-populated, such as time, location, cost, etc., so the sparsity is manageable. There's also a high priority on looking up things by virtually any dimension or combination of dimensions, so the array format is particularly useful.
I believe multiprocessing is a level above what's going on here. This is a VLIW (very long instruction word) processor. That means each instruction is made up of a number of shorter instruction segments that are handled separately, in parallel. Ideally, a CISC instruction will fan out into a number of sub-instructions which can be handled in one step by the chip, or interleaved with other instructions, so long as the net effect is identical to executing the same instructions in their original order. It's difficult to do this kind of translation and parallelizing/reordering without occasionally hitting an exception of some kind, such as a deadlock, which would indicate that the translation failed.
At which point one needs to back up to a safe point and try a different approach (possibly pulled in from the translation cache they've mentioned), until the instruction succeeds. The mechanism used to back out is a memory write journal with a commit/rollback facility.
"Deadlock" is a phenomenon which can occur in any system of multiple agents and shared resources. At the chip level the agents are executing instructions, and the resources being shared are registers, memory locations, and possibly cpu components such as fpu's and adders, etc. Process deadlock is a higher-level concept that applies to shared resources managed by an operating system.
I heard about something that translates SQL server stored procedures to Java for use in DB2's JVM. I'm not sure how much use that might be, but if DB2 is an option...
Try IBM's DB2 website; there's a mention of it there somewhere.
I couldn't make this up if I tried (from the Korea Herald):
Five nabbed on charges of spying for N.K.
The National Intelligence Service (NIS) announced yesterday that it caught five people who allegedly established a pro-North Korean underground revolutionary group in the South on Pyongyang's instructions and worked for the Communist regime.
The five former student activists were given secret names by North Korea, and were ordered by Pyongyang to spread North Korea's juche, or self-reliance, ideology among South Korean citizens in preparation for a "revolution in the South," the NIS said.
They sent reports to North Korea and received orders from Pyongyang through the Hotmail Web-based e-mail service, the NIS said.
Scene: Students lining up for 8 am body cavity search. A guard holds a scanner in one hand; the other hand is covered with a latex glove bearing a "George W. Bush for President" logo.
Student: "OUCh!"
Scanner: (Beep!)
Guard (peering at scanner): "What the....Where did you get a name like 'Pepsi Twelvepack'?"
There are a few aspects of remote admin that OS/2 does, and AFAIK NT doesn't.
1. The os/2 command shell talks to an ansi terminal (and likewise, shell windows are ansi terminals).
2. telnetd is built-in with the tcp/ip package.
3. DOS sessions were given the same ansi-terminal capability several years ago. I'm not sure how far this goes, but possibly full-screen text dos apps can run over telnet.
(slashdot is really slow today, so I'm retrying - sorry for dupes)
From the specs this turkey looks like a new version of the MWave DSP that IBM had a few years ago. The thing was supposed to be the greatest thing since sliced bread, with sound, modem, and fax support all in one. But IBM never got the drivers written for anything but Windoze, games never worked, and basically the market passed it by - all the marvelous flexibility of the DSP was wasted by incompetent marketing and software support. Now I see they have a separate sound chip, which is probably the only reason sound sort of works under linux, and I doubt if they could get the DSP people even to release the specs for 3rd party driver development.
I doubt if the total number of addresses would be exceeded, but it seems plausible that there would be situations where one would want the freedom to add bits to an existing addressing scheme, in order to gateway it to another network or address system. For instance, if you build a whizbang superfast neutrino network in the year 2028, and want to attach it to the existing internet, you could either (a) attach it through a gateway similar to ip_masquerade, where each packet is rewritten (try that at 80 Gbps, or (b) attach it directly to existing routers, with no address translation required, so long as the additional address bits can be appended to the end and forwarded intact. Variable length addressing would be helpful for (b).
A few years back I wrote a smart-ass response to somebody on Usenet, snickered quietly to myself, and then forgot about it. A few months later I saw someone mentioning a funny rant in Wired that sounded suspiciously similar to what I had posted. I went over to a newsstand, paid $4.95, and saw my rant in the letters section, authored by "anonymous". At no time did I recall every giving Wired permission to publish my incoherent ramblings, so I checked around, posted to one of the law newsgroups, and even flamed a few people in alt.wired. The basic facts I established were: I have until something like 75 years after my death to sue them; I don't need a copyright notice; damages are mainly limited to proven commercial value (none, I will freely admit, and I would probably have given them permission *if asked*), and, when you get right down to it, Wired is for ding-dongs. During that time period, Wired was attracting large numbers of zealots who thought that a pink-and-green dead trees publication was somehow revolutionizing the online world, and that things like copyrights, honesty, etc., were obsolete. One guy even seemed to think that Wired was actually an extension of Usenet, and therefore the magazine could freely profit from whatever it could grab there. True, there are some questions as to where Usenet actually ends, since, even more than the web, it's a distributed, multi-copying system, but I'm sure most would agree it stops somewhere short of overpriced, advertisement-laden lifestyle magazines sitting in traditional newsracks. In any case I have something fun to do in my old age, should I ever get the desire to sue them.
A little history: Apple put $100M into Samsung a few months ago to build their flat-panel assembly lines, and I seem to recall seeing something about a 22" flat panel from them a week or so ago. Apple's gambit is apparently to monopolize the supply of these things, since most of Northeast Asia is scrambling to build factories to produce them.
I downloaded the source and found it contains 2798 entries. When I pipe the list through sort -u, I find only 1399 of these filenames are unique. Is there an echo in here...in here? (I guess that means some files have been GPL'd *twice*.)
I think it refers to the physical layout of the bits in the material, and the amount of "empty space" between them. There are also things to help locate the beginning of each block as it passes under the head, and generally to reduce positional error, which may be all that they mean here.
Yeah, I'm about at my 10-year anniversary in SF, and it's a pretty good example of "Old Urbanism". So-called "conservatives" can rant on about how subsidizing roads, gas, and sprawl is not "big government", but I know better.
In accordance with a secret government plan, on January 1, 1900, the all computers will switch from measuring data in kilobytes, units of 1000 bytes, to kibibytes, units of 1024 bytes. In the resulting confusion, computers worldwide will overflow their data partitions, causing bank failures, mass panics, child TV viewing, and venereal disease outbreaks which will end society as we know it.
Here is a story about the stunt from the Korea Times, with a mention of you-know-who. Darn it, there must be a lot of nerds in Korea.
I also spotted this article about a "Hacker's Lab" that allows crackers to work their way up to something like a "black belt" in cracking, by undertaking a series of canned cracks. It might be cool, might be lame, but it's kind of funny.
I'm sure the Russians thought about using supercomputers to model bread lines before they gave up and decided to let market forces take over. Here's my prediction, based on empirical observation, which can be a remarkable tool for predicting the output of computer models: if you tax people to build something, and then give it away free, there will always be shortages.
Slashdot Mirror
SQL may not appear on the front end, but that may be what the server gets from the tool. Most OLAP queries aim at aggregating all the data within particular ranges, eg "all sales of books costing over $20, in stores in the Southwestern US, to customers who paid by credit card." An OLAP tool would build the query by presenting the user with choices about each dimension, and then accumulating SQL predicates to define each range, eg "price > 20", "payment_type = 'credit card'", etc. Once this process is over, the tool comes up with a whopper of a query, possibly involving dozens of predicates. The database then has to look at this beast, and without any human assistance, come up with a good way of fetching the data.
Multidimensional refers more to the physical data layout than the actual logical model, which can still be relational.
In a conventional RDBMS, a tuplet of values (a,b,c,d), say, can end up in any free block in the table (and indexes are needed for each column searched). In a multidimensional db, tuplets are laid out as if their values were array subscripts, the way a multidimensional array would be in, say, FORTRAN. It's a good deal easier to find data that way, since it's the same as an array lookup. The downside is that the data has to be fairly dense, or else you'll end up with a lot of empty holes and wasted space.
In a data warehouse, the array is usually organized along dimensions that are well-populated, such as time, location, cost, etc., so the sparsity is manageable. There's also a high priority on looking up things by virtually any dimension or combination of dimensions, so the array format is particularly useful.
I believe multiprocessing is a level above what's going on here. This is a VLIW (very long instruction word) processor. That means each instruction is made up of a number of shorter instruction segments that are handled separately, in parallel. Ideally, a CISC instruction will fan out into a number of sub-instructions which can be handled in one step by the chip, or interleaved with other instructions, so long as the net effect is identical to executing the same instructions in their original order. It's difficult to do this kind of translation and parallelizing/reordering without occasionally hitting an exception of some kind, such as a deadlock, which would indicate that the translation failed.
At which point one needs to back up to a safe point and try a different approach (possibly pulled in from the translation cache they've mentioned), until the instruction succeeds. The mechanism used to back out is a memory write journal with a commit/rollback facility.
"Deadlock" is a phenomenon which can occur in any system of multiple agents and shared resources. At the chip level the agents are executing instructions, and the resources being shared are registers, memory locations, and possibly cpu components such as fpu's and adders, etc. Process deadlock is a higher-level concept that applies to shared resources managed by an operating system.
Try IBM's DB2 website; there's a mention of it there somewhere.
Five nabbed on charges of spying for N.K.
The National Intelligence Service (NIS) announced yesterday that it
caught five people who allegedly established a pro-North Korean
underground revolutionary group in the South on Pyongyang's
instructions and worked for the Communist regime.
The five former student activists were given secret names by North
Korea, and were ordered by Pyongyang to spread North Korea's juche, or
self-reliance, ideology among South Korean citizens in preparation for
a "revolution in the South," the NIS said.
They sent reports to North Korea and received orders from Pyongyang
through the Hotmail Web-based e-mail service, the NIS said.
Scene: Students lining up for 8 am body cavity search. A guard holds a scanner in one hand; the other hand is covered with a latex glove bearing a "George W. Bush for President" logo.
Student: "OUCh!"
Scanner: (Beep!)
Guard (peering at scanner): "What the....Where did you get a name like 'Pepsi Twelvepack'?"
jonkatz.com is taken!
Yeah, and phones will only need two buttons!
Sorry.
1. The os/2 command shell talks to an ansi terminal (and likewise, shell windows are ansi terminals).
2. telnetd is built-in with the tcp/ip package.
3. DOS sessions were given the same ansi-terminal capability several years ago. I'm not sure how far this goes, but possibly full-screen text dos apps can run over telnet.
(slashdot is really slow today, so I'm retrying - sorry for dupes)
If the flying part doesn't work out, the company is poised to take over the CPU cooling market.
Caveat Emptor!
I doubt if the total number of addresses would be exceeded, but it seems plausible that there would be situations where one would want the freedom to add bits to an existing addressing scheme, in order to gateway it to another network or address system.
For instance, if you build a whizbang superfast neutrino network in the year 2028, and want to attach it to the existing internet, you could either (a) attach it through a gateway similar to ip_masquerade, where each packet is rewritten (try that at 80 Gbps, or (b) attach it directly to existing routers, with no address translation required, so long as the additional address bits can be appended to the end and forwarded intact. Variable length addressing would be helpful for (b).
A few years back I wrote a smart-ass response to somebody on Usenet, snickered quietly to myself, and then forgot about it. A few months later I saw someone mentioning a funny rant in Wired that sounded suspiciously similar to what I had posted. I went over to a newsstand, paid $4.95, and saw my rant in the letters section, authored by "anonymous". At no time did I recall every giving Wired permission to publish my incoherent ramblings, so I checked around, posted to one of the law newsgroups, and even flamed a few people in alt.wired. The basic facts I established were: I have until something like 75 years after my death to sue them; I don't need a copyright notice; damages are mainly limited to proven commercial value (none, I will freely admit, and I would probably have given them permission *if asked*), and, when you get right down to it, Wired is for ding-dongs. During that time period, Wired was attracting large numbers of zealots who thought that a pink-and-green dead trees publication was somehow revolutionizing the online world, and that things like copyrights, honesty, etc., were obsolete. One guy even seemed to think that Wired was actually an extension of Usenet, and therefore the magazine could freely profit from whatever it could grab there. True, there are some questions as to where Usenet actually ends, since, even more than the web, it's a distributed, multi-copying system, but I'm sure most would agree it stops somewhere short of overpriced, advertisement-laden lifestyle magazines sitting in traditional newsracks. In any case I have something fun to do in my old age, should I ever get the desire to sue them.
A little history: Apple put $100M into Samsung a few months ago to build their flat-panel assembly lines, and I seem to recall seeing something about a 22" flat panel from them a week or so ago. Apple's gambit is apparently to monopolize the supply of these things, since most of Northeast Asia is scrambling to build factories to produce them.
I downloaded the source and found it contains 2798 entries. When I pipe the list through sort -u, I find only 1399 of these filenames are unique. Is there an echo in here...in here? (I guess that means some files have been GPL'd *twice*.)
I think it refers to the physical layout of the bits in the material, and the amount of "empty space" between them. There are also things to help locate the beginning of each block as it passes under the head, and generally to reduce positional error, which may be all that they mean here.
Yeah, I'm about at my 10-year anniversary in SF, and it's a pretty good example of "Old Urbanism". So-called "conservatives" can rant on about how subsidizing roads, gas, and sprawl is not "big government", but I know better.
Jason Priestley? Why don't they put his picture on the padlock cover?
In this universe black holes suck; in another universe, they rule!
Try info@joker.com. They got back to me fairly quickly, in English, when I had a problem.
In accordance with a secret government plan, on January 1, 1900, the all computers will switch from measuring data in kilobytes, units of 1000 bytes, to kibibytes, units of 1024 bytes. In the resulting confusion, computers worldwide will overflow their data partitions, causing bank failures, mass panics, child TV viewing, and venereal disease outbreaks which will end society as we know it.
I also spotted this article about a "Hacker's Lab" that allows crackers to work their way up to something like a "black belt" in cracking, by undertaking a series of canned cracks. It might be cool, might be lame, but it's kind of funny.
Sounds like you should get what the motorists got: a $140/month raise.
I'm sure the Russians thought about using supercomputers to model bread lines before they gave up and decided to let market forces take over. Here's my prediction, based on empirical observation, which can be a remarkable tool for predicting the output of computer models: if you tax people to build something, and then give it away free, there will always be shortages.