Do some searches for "recursive sql". It's in the standard, but nobody put together a full implementation, unless something's come out in the last few years when I stopped following the issue. Briefly, it's an extension of SQL to allow selection of recursive sets, i.e. sets which refer to themselves. An example is "the set consisting of the root comment, union the comments whose parents are in the set" to get the full comment hierarchy.
In general this feature should allow traversal of any connected graph, with cycles, etc. Oracle's hack only works on trees, and fails on cycles, last time I checked.
There's no way to dislodge the big subsidies, so "feel good" rebates are about the only way to go. Unfortunately these just beget new unfair subsidy schemes.
I ride a bike to work, but I don't qualify for the "Low Emission Vehicle" tax credit. I love Big Brother!
There are actually standard designations for how a fluorescent light starts. The classic flicker-on is called a preheat fixture, since the filaments on each end of the bulb have to warm up like incandescents for a second or two before lighting with an inductive jolt. "Rapid start" are kept preheated with a small current (seems like a waste), and "instant start" ballasts use a high-voltage jolt to start the arc in a few milliseconds. Instant start usually means an electronic, i.e. better, non-buzzing, high frequency ballast.
If you look carefully at a CFL package, it will usually say "instant start" if it has that feature, or "Ikea" if it doesn't:-).
Does anybody know what the name "Winchester" refers to in this case?
My father worked on this product in the late 60's and beyond, but I never figured out if they were referring to the Mystery House, the boulevard, or the repeating rifle (all from the same family). My guess was that the Mystery House was chosen as a precursor to San Jose Building 5:-).
The real Google has duplicate detection to handle situations like these. Their crawl ends up with something like 30% duplicates from different sources, things like the same online manual repeated on dozens of different sites, mirrors, multiple servers, etc. They use various approximate-matching algorithms to find near-duplicates and merge them, so that the search results don't show the same document with a hundred different urls.
Unfortunately feature holes like this are why the thing hasn't taken off. If I have to submit a list of urls to avoid duplication, I might as well index the stuff myself.
It's been a long time coming, but we RT-11 users are finally getting revenge for that crappy GUI, 32k memory limit, and hype-driven "plug and pray" jumper architecture. It's time to say goodbye to proprietary "ODT-ready" keyboards and a system that locks up even doing simple tasks, like loading a CSR without disabling interrupts.
What's surprising about this? Looking at CS, I've wondered what people did in the past with new technologies. Did ancient Rome have plumbing professors? Did Oxford have a Cotton-milling Science department in the 1800's? I can just imagine the help-wanted ads for "Rock Star" toilet mechanics way back when.
There were periods when railroads, electricity, steam engines, bicycles, and the like were considered paradigm-changing technologies, worthy of study at the highest level, but they've gradually settled down into a few hardcore academic subjects (eg Mechanical Engineering), and a wide range of skilled trades (Mechanic, Electrician, Plumber). As the pace of innovation slows, the need for on-the-spot engineering diminishes, and the main skill is in applying standards and more static design principles, in the most efficient (labor-wise) and reliable way possible.
My guess is that, in the worst case, CS will be gradually whittled down until it has only a few active areas of research, and end up merged with EE as many colleges already have. Perhaps someday both will disappear into a more general engineering discipline, or will be dropped altogether.
The industry spent billions hedging against the small possibility of a Y2K failure, and then outsourced the same applications to a security nightmare like this one. I wonder what "Y2010 Certified" will mean -- heavily vetted personnel, most likely.
I had the same thought the other day, while killing off a bunch of popups from some Indian newspaper that was displayed first on Google News (looks like some kind of spam scheme).
Ideally, I would disallow Javascript on an url the first time it's loaded, and prompt the user on whether to allow it for that page. Thereafter the setting could be remembered for the url.
I keep turning J-script off while browsing, and then going back to a page I'm developing and waiting for it to respond. I swear it takes me several minutes to remember to turn it back on.
As a matter of fact I dug out my theory of economic froth at dinner the other day -- the idea that things like the web boom are symptoms of excess wealth that can be put to no useful purpose.
It's similar to my posts on slashdot where...oops, I'll explain later.
I've seen some Hollywood movies, so I can guarantee that when they collide, the iceberg will fly up into the air, flip over, and explode in an immense ball of flame.
Re:Cool! Just like form AutoComplete
on
Google Suggest
·
· Score: 1
It looks like a simple prefix match on a trie, with some pruning to get the top 10 results. There's no approximate matching or spelling correction which would make it harder.
So for a string of length k, we have:
1. walk down k nodes from the root to a node S -- (O(k)).
2. Explore leaf-node descendants of S, retrieving those with the top 10 counts. Each leaf represents a complete string in the trie, and I'll assume no strings are prefixes of others.
Step 2 could be done in a number of ways. I suppose each node could be labelled with the maximum count of any leaf node below it, to allow easy navigation to the highest-count descendant in O(1) (or O(alphabetsize) ) per level. Finding the second-ranked, third, etc. would then be a matter of deleting the first-ranked leaf (temporarily) and finding the top-ranked descendant again in the resulting trie, deleting it, and so on, iterating until 10 have been found.
Overall step 2 would be about 10 * (the average depth of a leaf, minus k) * O(1), or O( avg string length - k ).
Re:Implementation details
on
Google Suggest
·
· Score: 2, Interesting
I was thinking it would be nice if my iBook could auto-recognize my home hub and auto-login there. Outside the "zone", it could default back to login-on-wake. Coupled with the various options for encrypting my home directory, etc., this could be similar to what Woz is proposing.
Of course, I haven't sat down to figure out the cryptological protocol needed. For the average thief, a simple insecure method would probably work.
Slashdot Mirror
Do some searches for "recursive sql". It's in the standard, but nobody put together a full implementation, unless something's come out in the last few years when I stopped following the issue. Briefly, it's an extension of SQL to allow selection of recursive sets, i.e. sets which refer to themselves. An example is "the set consisting of the root comment, union the comments whose parents are in the set" to get the full comment hierarchy.
In general this feature should allow traversal of any connected graph, with cycles, etc. Oracle's hack only works on trees, and fails on cycles, last time I checked.
There's no way to dislodge the big subsidies, so "feel good" rebates are about the only way to go. Unfortunately these just beget new unfair subsidy schemes.
I ride a bike to work, but I don't qualify for the "Low Emission Vehicle" tax credit. I love Big Brother!
A GE Starcoat F32T8, available in most hardware stores, gets 2650 lumens, or 82.8 lumens/watt.
There are actually standard designations for how a fluorescent light starts. The classic flicker-on is called a preheat fixture, since the filaments on each end of the bulb have to warm up like incandescents for a second or two before lighting with an inductive jolt. "Rapid start" are kept preheated with a small current (seems like a waste), and "instant start" ballasts use a high-voltage jolt to start the arc in a few milliseconds. Instant start usually means an electronic, i.e. better, non-buzzing, high frequency ballast.
:-).
If you look carefully at a CFL package, it will usually say "instant start" if it has that feature, or "Ikea" if it doesn't
Does anybody know what the name "Winchester" refers to in this case?
:-).
My father worked on this product in the late 60's and beyond, but I never figured out if they were referring to the Mystery House, the boulevard, or the repeating rifle (all from the same family). My guess was that the Mystery House was chosen as a precursor to San Jose Building 5
But this guy really has me wondering. What would Math smell like?
I used to have a Real Analysis text that smelled like Garlic Shrimp, but only because I spilled some on it.
$ host -a gmail.co.uk ;; ->>HEADER- opcode: QUERY, status: NOERROR, id: 55781 ;; flags: qr rd ra; QUERY: 1, ANSWER: 2, AUTHORITY: 2, ADDITIONAL: 2 ;; QUESTION SECTION: ;gmail.co.uk. IN ANY ;; ANSWER SECTION: ;; AUTHORITY SECTION: ;; ADDITIONAL SECTION:
Trying "gmail.co.uk"
gmail.co.uk. 172800 IN NS ns0.geneticsnet.co.uk.
gmail.co.uk. 172800 IN NS ns1.geneticsnet.co.uk.
gmail.co.uk. 172800 IN NS ns1.geneticsnet.co.uk.
gmail.co.uk. 172800 IN NS ns0.geneticsnet.co.uk.
ns0.geneticsnet.co.uk. 172800 IN A 81.31.106.200
ns1.geneticsnet.co.uk. 172800 IN A 81.31.106.201
This has been the same since gmail was announced. Google didn't even bother to check if gmail was available in all TLD's.
At least they tested the functionality...right?
Search isn't about search, it's about advertising.
The ads on the Google page are put there by Google, not Anandtech. I assume Anand will have search-targeted ads on their own results page soon enough.
The real Google has duplicate detection to handle situations like these. Their crawl ends up with something like 30% duplicates from different sources, things like the same online manual repeated on dozens of different sites, mirrors, multiple servers, etc. They use various approximate-matching algorithms to find near-duplicates and merge them, so that the search results don't show the same document with a hundred different urls.
Unfortunately feature holes like this are why the thing hasn't taken off. If I have to submit a list of urls to avoid duplication, I might as well index the stuff myself.
I once spent a summer figuring out how to write device interfaces by trial and error. ODT was my closest friend.
Alas, I think slashdot is going the way of the PDP-11. I post a joke in response to a story about nothing, and I get marked "off-topic".
It's been a long time coming, but we RT-11 users are finally getting revenge for that crappy GUI, 32k memory limit, and hype-driven "plug and pray" jumper architecture. It's time to say goodbye to proprietary "ODT-ready" keyboards and a system that locks up even doing simple tasks, like loading a CSR without disabling interrupts.
Because Google has duplicate detection technology.
The practice of hyping up every single event involving Google, or, in the absence of news, slight changes in air pressure.
What's surprising about this? Looking at CS, I've wondered what people did in the past with new technologies. Did ancient Rome have plumbing professors? Did Oxford have a Cotton-milling Science department in the 1800's? I can just imagine the help-wanted ads for "Rock Star" toilet mechanics way back when.
There were periods when railroads, electricity, steam engines, bicycles, and the like were considered paradigm-changing technologies, worthy of study at the highest level, but they've gradually settled down into a few hardcore academic subjects (eg Mechanical Engineering), and a wide range of skilled trades (Mechanic, Electrician, Plumber). As the pace of innovation slows, the need for on-the-spot engineering diminishes, and the main skill is in applying standards and more static design principles, in the most efficient (labor-wise) and reliable way possible.
My guess is that, in the worst case, CS will be gradually whittled down until it has only a few active areas of research, and end up merged with EE as many colleges already have. Perhaps someday both will disappear into a more general engineering discipline, or will be dropped altogether.
The industry spent billions hedging against the small possibility of a Y2K failure, and then outsourced the same applications to a security nightmare like this one. I wonder what "Y2010 Certified" will mean -- heavily vetted personnel, most likely.
I had the same thought the other day, while killing off a bunch of popups from some Indian newspaper that was displayed first on Google News (looks like some kind of spam scheme).
Ideally, I would disallow Javascript on an url the first time it's loaded, and prompt the user on whether to allow it for that page. Thereafter the setting could be remembered for the url.
I keep turning J-script off while browsing, and then going back to a page I'm developing and waiting for it to respond. I swear it takes me several minutes to remember to turn it back on.
for( unsigned i = 10; i >= 0; i-- );
A fool and his money are soon venture capital.
-- someone's sig.
As a matter of fact I dug out my theory of economic froth at dinner the other day -- the idea that things like the web boom are symptoms of excess wealth that can be put to no useful purpose.
It's similar to my posts on slashdot where...oops, I'll explain later.
I've seen some Hollywood movies, so I can guarantee that when they collide, the iceberg will fly up into the air, flip over, and explode in an immense ball of flame.
People call this "full text search", but it's really "lexical unit search", i.e. only on the terms which are recognized by the parser before indexing.
It's the Google Cool Campus now.
It looks like a simple prefix match on a trie, with some pruning to get the top 10 results. There's no approximate matching or spelling correction which would make it harder.
So for a string of length k, we have:
1. walk down k nodes from the root to a node S -- (O(k)).
2. Explore leaf-node descendants of S, retrieving those with the top 10 counts. Each leaf represents a complete string in the trie, and I'll assume no strings are prefixes of others.
Step 2 could be done in a number of ways. I suppose each node could be labelled with the maximum count of any leaf node below it, to allow easy navigation to the highest-count descendant in O(1) (or O(alphabetsize) ) per level. Finding the second-ranked, third, etc. would then be a matter of deleting the first-ranked leaf (temporarily) and finding the top-ranked descendant again in the resulting trie, deleting it, and so on, iterating until 10 have been found.
Overall step 2 would be about 10 * (the average depth of a leaf, minus k) * O(1), or O( avg string length - k ).
http://www.google.com/complete/search?hl=en&js=tru e&qu=chicken.
You can substitute any string for "chicken", and hl is the language.
I was thinking it would be nice if my iBook could auto-recognize my home hub and auto-login there. Outside the "zone", it could default back to login-on-wake. Coupled with the various options for encrypting my home directory, etc., this could be similar to what Woz is proposing.
Of course, I haven't sat down to figure out the cryptological protocol needed. For the average thief, a simple insecure method would probably work.