So if I understand this correctly, after railing about how Google was leeching off of others without paying a dime, Hinton is now going to use Google for his own profit without paying Google a dime...
This is how Obama got elected. People who have absolutely no clue how the economy works voting.
On the plus side, apparently a large proportion of the population understood how an improvement on foreign relations could be achieved. Even if Palin had and could see Russia (oh, man, one of the best jokes ever)
But seriously, Bush understood how the economy works? How did we end up in this mess then?
This would be rare, but if you add a node to the rightmost side of the tree, then the only update you need to do is update the rgt column value of the root node.
All its ancestors rgt values change, so there is only one case where no other nodes are affected: inserting the root in the empty tree.
Thanks for a constructive discussion, a actually learned something on/. today:-)
And in both cases they build their cloud solutions on open source. Amazon EC2 offered Linux first and then Windows at a 25% higher cost. So, isn't Open Source competing very well?
UPDATE trees SET rgt = rgt + 27, lft = lft + 27 WHERE lft > 107 AND tree_id = 34
I don't know much about how databases write physical data, but I assumed that an update to a single row, regardless of how many columns are updated, counts as a single update.
I don't know what you mean with the tree_id, but you have to update all nodes that follow the inserted one, so on average, half the nodes in the tree or half the rows in that table. According to your link 1
UPDATE nested_category SET rgt = rgt + 2 WHERE rgt > @myLeft;
UPDATE nested_category SET lft = lft + 2 WHERE lft > @myLeft;
E.g. if you take the image with the venn diagrams, and insert a new child "ipod" in "mp3", after "flash" these are the changes:
Electronics rgt = 22
Portable Electronics rgt = 21
MP3 rgt = 16
CD rgt = 18, lft = 17
2 Way R. rgt = 20, lft = 19
Insert iPod, rgt = 14, lft = 15
5 of the 10 nodes are updated, and another 3 are updated again. I don't thinl there is any situation where only one node is affected.
2 UPDATES, you forgot lft. And since the tree is bludgeoned into the square table, yes you have to 'traverse' the tree. That UPDATE accesses each row that is affected. My first impression was that every node/row was accessed.
How big are your trees, and how many concurrent users do you want updating them?
My conclusion was: OK if update/insert is rare. Not OK when large tree, frequent structural modification, many concurrent users. That I've had to work with very large trees is another story, but it was about billing customers while operations continue, SQLServer stored procedures.
It's an intelligent hack, but it is not proof that RDBMS/SQL is a best fit for trees.
So in order to insert one node you have to traverse the entire tree to recalculate left/right values? that is, update each row (either left or right value)? That kind of rules it out when there are lots of concurrent users modifying the tree.
If your data isn't complex enough to require a RDBMS, you almost certainly don't need a program.
Really? IM, word processor, spreadsheet, vector graphics, photo editor,... Google probably uses MapReduce without a "normal" RDBMS behind it - is that data complex enough?
I really hate SQL. It failed at being a natural language search tool for business people. Others think this too: The Third Manifesto
With that out of the way, you mention the point-in-time-recovery and that touches on another issue I profoundly dislike with databases. It is possible to go back to a previous state, but then often applications depending on the DB schema break. When managing software versioning, one of the most difficult parts to get right, one of the highest risks in breaking things, is keeping track of the database schema changes.
I agree with you that the OP doesnt seem to have thought this through.
It took more than a decade to remove presentation from content in HTML, and now moving that back in?
If the platform independent formatting is so important, why not use PDF? That is also very well supported on different platforms. And it is reasonably easy to generate with XML + XSL-FO.
Why 1 HTML page to x Paper pages? Why not x HTML pages?
Why not use XML + CSS? most browsers support this.
Why not use SVG? most browsers support this too.
I think the ODF way is still the best for archiving, keeping content and markup data separated (separate XML) and together (single ZIP). And it is an open standard with freely available software - not just OpenOffice.org.
There is a Single OpenDocument XML File variant for ODF, where everything is a single XML file (root tag office:document) if it really has to be.
Slashdot Mirror
So if I understand this correctly, after railing about how Google was leeching off of others without paying a dime, Hinton is now going to use Google for his own profit without paying Google a dime...
Yeah, it's payback time! Kind of.
This is how Obama got elected. People who have absolutely no clue how the economy works voting.
On the plus side, apparently a large proportion of the population understood how an improvement on foreign relations could be achieved. Even if Palin had and could see Russia (oh, man, one of the best jokes ever)
But seriously, Bush understood how the economy works? How did we end up in this mess then?
Wait I didn't realize slashdot was primarily news, the articles are just fluff. The comments are always a better read if you want to read a story.
Pay attention, it's at the top of each page: Slashdot. News for Herds. Comments that Matter. Or something. Can't be bothered to scroll up to read it.
So basically you argue that:
Again, the solution isn't more government regulation (which also has loophoes), but less (none!).
Look at what happened to failed states like Somalia and Sudan. Warlords. Pirates. Al-Qaeda.
[...]since most financial institutions [...]
I know, because I work for $LARGE_BANK, and we use it there. [...]
That's a hell of an extrapolation. It could only be correct if there are less than 2.0 possible $LARGE_BANK values...
I have absolutely no idea, but you are so wrong.
Plus the understanding of perspective drawing was very poor prior to 1700ish
...and so the current generation builds on the knowledge and experience gathered by their predecessors.
[sarcasm]Great thing we now have "Intellectual Property" and virtually unlimited copyright, DRM, DMCA ... to drive innovation[/sarcasm]
But well, this is /. after all where we argue about pointless points (myself included).
I must disagree with that
At least for some it is free.
This would be rare, but if you add a node to the rightmost side of the tree, then the only update you need to do is update the rgt column value of the root node.
All its ancestors rgt values change, so there is only one case where no other nodes are affected: inserting the root in the empty tree.
Thanks for a constructive discussion, a actually learned something on /. today :-)
-i think the two of -us differ in opinion. Look it up, it's viruses.
You're a wise person.
And in both cases they build their cloud solutions on open source. Amazon EC2 offered Linux first and then Windows at a 25% higher cost. So, isn't Open Source competing very well?
Let me wikipedia that for you. Tupolev and Boeing. The Tupolev crew followed ATC and disregarded instrument.
So in this scenario, if the Tupolev would follow the Airbus philosophy:
Airbus scenario + idiot pilot = no crash
Airbus scenario + smart pilot = no crash
Now, even Airbus doesn't design aircraft like that. Fly-by wire controls are designed like that, not instruments and indicators.
UPDATE trees SET rgt = rgt + 27, lft = lft + 27 WHERE lft > 107 AND tree_id = 34
I don't know much about how databases write physical data, but I assumed that an update to a single row, regardless of how many columns are updated, counts as a single update.
I don't know what you mean with the tree_id, but you have to update all nodes that follow the inserted one, so on average, half the nodes in the tree or half the rows in that table. According to your link 1
UPDATE nested_category SET rgt = rgt + 2 WHERE rgt > @myLeft; UPDATE nested_category SET lft = lft + 2 WHERE lft > @myLeft;
E.g. if you take the image with the venn diagrams, and insert a new child "ipod" in "mp3", after "flash" these are the changes:
5 of the 10 nodes are updated, and another 3 are updated again. I don't thinl there is any situation where only one node is affected.
2 UPDATES, you forgot lft. And since the tree is bludgeoned into the square table, yes you have to 'traverse' the tree. That UPDATE accesses each row that is affected. My first impression was that every node/row was accessed.
How big are your trees, and how many concurrent users do you want updating them?
My conclusion was: OK if update/insert is rare. Not OK when large tree, frequent structural modification, many concurrent users. That I've had to work with very large trees is another story, but it was about billing customers while operations continue, SQLServer stored procedures.
It's an intelligent hack, but it is not proof that RDBMS/SQL is a best fit for trees.
The only plausible explanation is that Microsoft must have bought Apple...
Thanks for the information.
So in order to insert one node you have to traverse the entire tree to recalculate left/right values? that is, update each row (either left or right value)? That kind of rules it out when there are lots of concurrent users modifying the tree.
database generated id
If your data isn't complex enough to require a RDBMS, you almost certainly don't need a program.
Really? IM, word processor, spreadsheet, vector graphics, photo editor, ... Google probably uses MapReduce without a "normal" RDBMS behind it - is that data complex enough?
Oh. Forgot tree structures (or anything not tabular) and retrieving them.
I really hate SQL. It failed at being a natural language search tool for business people. Others think this too: The Third Manifesto
With that out of the way, you mention the point-in-time-recovery and that touches on another issue I profoundly dislike with databases. It is possible to go back to a previous state, but then often applications depending on the DB schema break. When managing software versioning, one of the most difficult parts to get right, one of the highest risks in breaking things, is keeping track of the database schema changes.
I agree with you that the OP doesnt seem to have thought this through.
I think the ODF way is still the best for archiving, keeping content and markup data separated (separate XML) and together (single ZIP). And it is an open standard with freely available software - not just OpenOffice.org.
There is a Single OpenDocument XML File variant for ODF, where everything is a single XML file (root tag office:document) if it really has to be.
or a throwaway install on a USB stick...
Seriously: " who direct them to sites hosting keyloggers & trojans.", not much of social engineering - if your basic setup is secure, it wouldn't work
And that was the one time Albert was wrong...