The "old" row cannot be deleted at transaction commit! Other transactions might still be reading it.
That's why PostgreSQL has VACUUM.
Moreover, VACUUM by itself just marks a row in the Free Space Manager (FSM) as free and writes over it at the next opportunity. VACUUM FULL will actually shrink the size of the table if that's what you need.
VACUUM should be automatic though, and eventually will be.
in mssql, i can alter a table definition that has views defined against it, and if the resulting set is still consistent with the dependent views, no problem. in postgresql, i have to *delete* all the views (and i believe some other objects) that depend on the table, alter the table, and then recreate the dependent objects. BIG hassle, especially while you're in the process of developing an application and the data definition changes.
Can you show what you mean? In PostgreSQL 8.0.0beta1, the following works just fine:
test=# create table t1(i int); CREATE TABLE test=# create view v1 as select * from t1; CREATE VIEW test=# alter table t1 add column j int; ALTER TABLE test=# insert into t1 values(1,11); INSERT 17234 1 test=# select * from v1;
i ---
1 (1 row)
So what's the problem? That's how I'd expect it to behave.
communism as a governmental type, though. It's an ideal system
I wouldn't say it's ideal. Let's say you ignore the fact that communism generally means slavery, there still exist major economic problems.
Communism is a centrally planned economy. Leaders (perhaps democratically elected) plan what someone else produces, and how that product is distributed. It's too much micromanagement and heirarchy.
Unfortunately, central planning (even through a sophisticated heirarchy of decision making) doesn't work very effectively on a large scale for complex systems. Usually there's some element of central planning, but effective large-scale planners don't micromanage (even through a heirarchy).
These aren't just meaningless details. They are very relevent today. Politicians speak about rationing out health care or hiding the true costs of health care. They don't use those words, they speak about "right to medical care" or some such nonsense.
We are seeing problems with this already in insurance companies. It became a problem when employers started providing health plans, and employees using the health plans. The employee is so far-removed from the actual costs that there is no reason for the patient to not demand an MRI and a heart surgeon to fix their sprained ankle.
In a capitalistic transaction, the person paying the costs is the same person as the person recieving the benefit and also the same person making the decision. This is becoming less and less true in U.S. healthcare.
Well, big surprise, there are wait lists for surgery, wait lists for doctors, insurance companies denying coverage, huge costs for doctors' time, and "insurance" that covers things you know you'll have to deal with no matter what (Are childbirth and childrens vaccinations really all that surprising? It's not as if you're buying meteor insurance. You're better off saving the money and paying yourself, so the insurance companies don't profit.).
Yeah, WalMart's amazing. Just as long as you don't work for them.
Wal-Mart is the largest employer in the U.S. I suppose they could fire 50% of their workforce and give a few more benefits to their workers. To increase per-employee costs even a small amount means killing the razor thin margins they have, and would require closing a large portion of their stores.
I wouldn't advise working there if you have a choice, but sometimes you have to recognize the reality. If Wal-Mart behaved like [insert company you think is a better employer here], than it would only be able to support as many employees as that company you think is a better employer.
Then, what about all those people that no longer work for wal-mart? They end up choosing an even less desirable alternative. I shudder to think what that is.
You don't even need to go into that to show the problem with his sig.
Here is the text of the 2nd: A well regulated Militia, being necessary to the security of a free State, the right of the people to keep and bear Arms, shall not be infringed.
Now, allow me to change just a few words: A well educated electorate, being necessary to the security of a free State, the right of the people to keep and read books, shall not be infringed.
Does the above mean that only the well-educated should be allowed to keep and read books? Of course not. The second clause stands on it's own. If the founding fathers had intended for only the militia to reserve their right to bear arms, they would have said "the right of the militia" not "the right of the people".
I can respect arguments that gun control laws are a good idea. I cannot respect arguments that gun control is constitutional.
PostgreSQL has a pretty darn good track record, and the only thing that can make it better is time and new installations, but your point is valid.
I would say that #2 is correct, except it has little or nothing to do with open source. It's just the fact the PostgreSQL is backed by small companies, and is unknown to many people. If you as an employee go with anyone other than the perceived market leader (in this case Oracle), than part of the responsibility for failure lies on you. So it has more to do with the market perception of a vendor. Anyone who chooses another product might be accused of "cheaping out" or something similar.
Consider Apache, the clear market leader. Nobody is accused of cheaping out when they download the source of apache for free.
There's no 'technical' difference between a first digit change upgrade and a second digit upgrade.
As I understood from reading the lists, there is a technical distinction. a.(b+1).c is supposed to be more stable than a.b.d, even though it may have some additional features.
8.0 added so many powerful features that the developers did not want to imply that it would be more stable than 7.4. 7.4 is pretty much rock-solid, and 8.0 might not achieve that reputation until 8.1 or so.
You don't have to wait for 8.0 to do that... works fine in 7.4 (and 7.3 if I remember correctly).
Returning result sets was a solved problem in PostgreSQL a long time ago. They're also referred to as "table functions" in the docs because you can select from them like a table.
It makes me wonder why PostgreSQL doesn't provide some syntax for stored procedures, just to avoid confusion that postgresql doesn't have them. As far as I can tell, pg has all the same functionality in functions.
The problem doesn't have much to do with who can access it. Once the information is there, our legal system will allow many ways to access it, even in civil proceedings, long before guilt is proven beyond a reasonable doubt.
The problem is the logging in the first place. What if you were required to carry a voice recorder wherever you went? What if you were required to have cameras always on inside your house?
Then, the moment anyone has probable cause against you, your 5th amendment rights are useless because you have already effectively testified against yourself.
Cars are a basic part of free life. You can't reasonably live without a car in most parts of the States, yet if you drive one you must give up many of your rights.
It happens with another important aspect of life also: school. It's the law: you have to send your kids to school. But when they go there, you turn your parental rights over to the school.
Eventually, the government will claim that the air belongs to the government, and in order to breathe it you implicity consent to searches. Is it that far from where we are now?
people at mysql should abandon their quest to make mysql into postgres
I haven't seen much evidence that MySQL wants to be like PostgreSQL.
My impression is that MySQL outright rejects many fundamental philosophies of PostgreSQL. For instance, in MySQL, it's legal to enter the date Feb 31st. I'm confident that the MySQL developers could fix that, but their philosophy is to put the responsibility into the application. MySQL also has the philosphy that if an error is encountered, keep going.
Contrast with PostgreSQL, which regidly enforces contraints and will most certainly give you an error if it encounters any problem or ambiguity. It also allows application developers to place more responsibility on the database.
The MySQL approach is perfectly valid. After all it's easier (read: cheaper) to stack up more application servers than to cluster a database.
However, I rarely see business rules effectively enforced in an application. Also, consider if you have multiple applications accessing the same database. If your data is inconsistant with your business rules, where is the bug? It could be in any of the applications, and it could be compounded throughout all of them as they see the invalid data. It's a debugging nightmare. Contrast with PostgreSQL: you just have constraints to enforce your business rules, and the app with the bug will encounter an error and will be stopped in its tracks before another application is affected.
(2) It has a new caching algorithm which should significantly improve the cache hit rate in most real-world situations. Specifically, they are changing from a "least-recently updated" algorithm (LRU) to a system that accounts for how often the page was accessed. In the old caching system, if a large table scan was done, the cache would be polluted and your small, frequently-accessed table wouldn't be in cache.
(3) point-in-time recovery implies incremental backups, which are also very valuable
Funny how MySQL releases even the client libraries as GPL instead of LGPL.
That means you can't even ship your non-GPL product with MySQL support unless you buy a commercial license from MySQL AB. Commercial databases such as Oracle don't even have this restriction!
My advice is to not kill women, and then you're pretty surely not going to be charged with a double murder.
If it was barely conceived, it would probably take an autopsy to even find out, and most likely it would just be a single count.
The "didn't know" argument is worthless. If a women was holding a baby and you shot and killed both it doesn't matter whether you thought the baby was just a shopping bag.
I would say any time after conception it's fair game for the double murder. The argument that "maybe it was going to die anyway" doesn't go very far with me either. It's sort of like if you kill someone at 40 saying "well, his life was about half over anyway, so really it's just a half murder".
The arguments for abortion rights usually revolve around the body belonging to the woman. Well, her body certainly doesn't belong to some 3rd party murderous psychopath.
If you drive, you concentrate on the road, and if you don't do that, you face the consequences when the inevitable happens.
I agree, but consider the implications.
We have a speed limit law because people felt that there should be some objective criteria for determining reckless behavior before an accident.
Your attitude is: "Do whatever you want until some harm is done, then we sort it out in court."
I think that's a reasonable position, but one problem is that often the harm can't be undone by any court. You can't deter someone from speeding if they die in the accident anyway (since that's pretty much the maximum deterrent already). You can't bring the other person back to life.
So, as with many laws there is a compromise. If an officer is given objective criteria to deter, he can deter you from the (supposedly) dangerous practice by citing you. The practice may or may not be dangerous in that particular case, but the officer doesn't really need to prove that, since the law lists the objective criteria itself as illegal.
Now, we run into conflict-of-interest problems when the fine is paid back to the same people who issue the ticket, but that's another issue.
Let's assume your comment is correct. Why do we listen to any countries on the U.N.? Heck, LIBYA is on the human rights committee.
So these people who say we should be "multilateral" are really saying that we should follow the advice of a bunch of dictators who don't let their people speak. "Multilateral" sounds a lot better until you start listing the actual countries out, huh?
I suppose we should follow those dictators, because if we ignored them we'd be imposing democracy on other people.
That's why the U.N. is a joke. It's a joke when they don't enforce their resolutions. It's a joke when Libya is on the human rights committee. It's a joke when we pretend that a dictator speaks for his people.
If Spain wants to join the cause, then encourage terrorism by giving the terrorists results ("Hey, let's make some more demands in Spain, they seem to buckle quickly"), that's their problem. They're a sovereign nation, after all. At least until the next group of tyrants occupies their country and the U.S. needs to bail 'em out again.
Regardless, the U.K. is hardly anything less than 1st world, and they are a strong U.S. ally.
Probably the best example of irony that I heard was: "It's ironic if you poke yourself in the eye putting on a pair of safety glasses".
When someone says that something is ironic, I always compare it to that. This thing about a slashdot editor supposedly being annoying and posting a story about annoying people is really nothing worth HAVING a word for. You could call him a hypocrite, or dense or something, but really nobody cares and I don't think a new word is necessary.
never mind that most of the taxes are paid by the rich anyways.
The more important question is so much simpler. Are taxes on the rich to low? To high? Just right? Most people who whine about "tax cuts for the rich" have no idea what taxes on the "rich" actually are, or who is considered "rich". If "tax cuts for the rich" are always bad, then we should just tax "rich" people at 100%. So, sometimes a tax cut is good, sometimes it's bad. It all depends on the numbers, and the associated economic implications (which affect everyone).
Few people have enough time to bother themselves with numbers, facts, or real economic analysis. It's so much easier to just buy into some rhetoric.
Of course, there could be legitimate concerns about said tax cut. I'm just critisizing the people that are taken in by the rhetoric without asking the obvious questions.
Let's say a program is licensed with BSD. The theory behind the GPL says that a corporation can make a closed version of the same software, but add in a bunch of "must-have" features and lure people to use their improved (but closed) version. Short-sighted people might be lured away and the open version would disappear.
But what are the realities? MySQL and PHP both used to be more free than they are now. MySQL used to have LGPL client libraries and GPL code, now the client libraries are GPL (which is so restrictive that many people need to buy commercial licenses from MySQL when all they want to do is add a MySQL DB driver to their application). PHP used to be GPL, now it's not.
I guess the moral of the story is that the most important thing is that the copyright is not held by any one entity. You can debate whether GPL is better than BSD, but it seems a moot point when one entity controls the entire copyright, because they can do tricky things by changing the license themselves (like mysql client libraries going LGPL -> GPL).
In that respect something like PostgreSQL is much safer and more likely to remain free because the copyright is held by so many people that nobody could change the license.
Slashdot Mirror
I would like to add also that some databases expect you to give planner hints or they preplan queries.
PostgreSQL uses up-to-date statistics to plan each query, meaning that when the data changes, the plans change also.
If you rely on planner hints and preplanned queries, then those hints or plans can become out of date, and slow you down.
The "old" row cannot be deleted at transaction commit! Other transactions might still be reading it.
That's why PostgreSQL has VACUUM.
Moreover, VACUUM by itself just marks a row in the Free Space Manager (FSM) as free and writes over it at the next opportunity. VACUUM FULL will actually shrink the size of the table if that's what you need.
VACUUM should be automatic though, and eventually will be.
Can you show what you mean? In PostgreSQL 8.0.0beta1, the following works just fine:
test=# create table t1(i int);
CREATE TABLE
test=# create view v1 as select * from t1;
CREATE VIEW
test=# alter table t1 add column j int;
ALTER TABLE
test=# insert into t1 values(1,11);
INSERT 17234 1
test=# select * from v1;
i
---
1
(1 row)
So what's the problem? That's how I'd expect it to behave.
http://www.agliodbs.com/
They also support PostgreSQL. Josh Berkus, a member of PostgreSQL core, owns it. Please correct me if I'm wrong.
Just to make this clear:
In order for the planner to do a good job, you must "VACUUM ANALYZE" regularly, not just "VACUUM". Or you could use "ANALYZE" by itself.
I would hope that 8.0 does a better job for you, I know they made some improvements to the planner.
communism as a governmental type, though. It's an ideal system
I wouldn't say it's ideal. Let's say you ignore the fact that communism generally means slavery, there still exist major economic problems.
Communism is a centrally planned economy. Leaders (perhaps democratically elected) plan what someone else produces, and how that product is distributed. It's too much micromanagement and heirarchy.
Unfortunately, central planning (even through a sophisticated heirarchy of decision making) doesn't work very effectively on a large scale for complex systems. Usually there's some element of central planning, but effective large-scale planners don't micromanage (even through a heirarchy).
These aren't just meaningless details. They are very relevent today. Politicians speak about rationing out health care or hiding the true costs of health care. They don't use those words, they speak about "right to medical care" or some such nonsense.
We are seeing problems with this already in insurance companies. It became a problem when employers started providing health plans, and employees using the health plans. The employee is so far-removed from the actual costs that there is no reason for the patient to not demand an MRI and a heart surgeon to fix their sprained ankle.
In a capitalistic transaction, the person paying the costs is the same person as the person recieving the benefit and also the same person making the decision. This is becoming less and less true in U.S. healthcare.
Well, big surprise, there are wait lists for surgery, wait lists for doctors, insurance companies denying coverage, huge costs for doctors' time, and "insurance" that covers things you know you'll have to deal with no matter what (Are childbirth and childrens vaccinations really all that surprising? It's not as if you're buying meteor insurance. You're better off saving the money and paying yourself, so the insurance companies don't profit.).
Yeah, WalMart's amazing. Just as long as you don't work for them.
Wal-Mart is the largest employer in the U.S. I suppose they could fire 50% of their workforce and give a few more benefits to their workers. To increase per-employee costs even a small amount means killing the razor thin margins they have, and would require closing a large portion of their stores.
I wouldn't advise working there if you have a choice, but sometimes you have to recognize the reality. If Wal-Mart behaved like [insert company you think is a better employer here], than it would only be able to support as many employees as that company you think is a better employer.
Then, what about all those people that no longer work for wal-mart? They end up choosing an even less desirable alternative. I shudder to think what that is.
You don't even need to go into that to show the problem with his sig.
Here is the text of the 2nd:
A well regulated Militia, being necessary to the security of a free State, the right of the people to keep and bear Arms, shall not be infringed.
Now, allow me to change just a few words:
A well educated electorate, being necessary to the security of a free State, the right of the people to keep and read books, shall not be infringed.
Does the above mean that only the well-educated should be allowed to keep and read books? Of course not. The second clause stands on it's own. If the founding fathers had intended for only the militia to reserve their right to bear arms, they would have said "the right of the militia" not "the right of the people".
I can respect arguments that gun control laws are a good idea. I cannot respect arguments that gun control is constitutional.
PostgreSQL has a pretty darn good track record, and the only thing that can make it better is time and new installations, but your point is valid.
I would say that #2 is correct, except it has little or nothing to do with open source. It's just the fact the PostgreSQL is backed by small companies, and is unknown to many people. If you as an employee go with anyone other than the perceived market leader (in this case Oracle), than part of the responsibility for failure lies on you. So it has more to do with the market perception of a vendor. Anyone who chooses another product might be accused of "cheaping out" or something similar.
Consider Apache, the clear market leader. Nobody is accused of cheaping out when they download the source of apache for free.
To which versions does that apply?
There's no 'technical' difference between a first digit change upgrade and a second digit upgrade.
As I understood from reading the lists, there is a technical distinction. a.(b+1).c is supposed to be more stable than a.b.d, even though it may have some additional features.
8.0 added so many powerful features that the developers did not want to imply that it would be more stable than 7.4. 7.4 is pretty much rock-solid, and 8.0 might not achieve that reputation until 8.1 or so.
You don't have to wait for 8.0 to do that... works fine in 7.4 (and 7.3 if I remember correctly).
Returning result sets was a solved problem in PostgreSQL a long time ago. They're also referred to as "table functions" in the docs because you can select from them like a table.
It makes me wonder why PostgreSQL doesn't provide some syntax for stored procedures, just to avoid confusion that postgresql doesn't have them. As far as I can tell, pg has all the same functionality in functions.
The problem doesn't have much to do with who can access it. Once the information is there, our legal system will allow many ways to access it, even in civil proceedings, long before guilt is proven beyond a reasonable doubt.
The problem is the logging in the first place. What if you were required to carry a voice recorder wherever you went? What if you were required to have cameras always on inside your house?
Then, the moment anyone has probable cause against you, your 5th amendment rights are useless because you have already effectively testified against yourself.
Cars are a basic part of free life. You can't reasonably live without a car in most parts of the States, yet if you drive one you must give up many of your rights.
It happens with another important aspect of life also: school. It's the law: you have to send your kids to school. But when they go there, you turn your parental rights over to the school.
Eventually, the government will claim that the air belongs to the government, and in order to breathe it you implicity consent to searches. Is it that far from where we are now?
people at mysql should abandon their quest to make mysql into postgres
I haven't seen much evidence that MySQL wants to be like PostgreSQL.
My impression is that MySQL outright rejects many fundamental philosophies of PostgreSQL. For instance, in MySQL, it's legal to enter the date Feb 31st. I'm confident that the MySQL developers could fix that, but their philosophy is to put the responsibility into the application. MySQL also has the philosphy that if an error is encountered, keep going.
Contrast with PostgreSQL, which regidly enforces contraints and will most certainly give you an error if it encounters any problem or ambiguity. It also allows application developers to place more responsibility on the database.
The MySQL approach is perfectly valid. After all it's easier (read: cheaper) to stack up more application servers than to cluster a database.
However, I rarely see business rules effectively enforced in an application. Also, consider if you have multiple applications accessing the same database. If your data is inconsistant with your business rules, where is the bug? It could be in any of the applications, and it could be compounded throughout all of them as they see the invalid data. It's a debugging nightmare. Contrast with PostgreSQL: you just have constraints to enforce your business rules, and the app with the bug will encounter an error and will be stopped in its tracks before another application is affected.
I'd also like to add a couple points:
(1) PostgreSQL will have a native win32 port
(2) It has a new caching algorithm which should significantly improve the cache hit rate in most real-world situations. Specifically, they are changing from a "least-recently updated" algorithm (LRU) to a system that accounts for how often the page was accessed. In the old caching system, if a large table scan was done, the cache would be polluted and your small, frequently-accessed table wouldn't be in cache.
(3) point-in-time recovery implies incremental backups, which are also very valuable
GPL instead of the BSD license
Funny how MySQL releases even the client libraries as GPL instead of LGPL.
That means you can't even ship your non-GPL product with MySQL support unless you buy a commercial license from MySQL AB. Commercial databases such as Oracle don't even have this restriction!
My advice is to not kill women, and then you're pretty surely not going to be charged with a double murder.
If it was barely conceived, it would probably take an autopsy to even find out, and most likely it would just be a single count.
The "didn't know" argument is worthless. If a women was holding a baby and you shot and killed both it doesn't matter whether you thought the baby was just a shopping bag.
I would say any time after conception it's fair game for the double murder. The argument that "maybe it was going to die anyway" doesn't go very far with me either. It's sort of like if you kill someone at 40 saying "well, his life was about half over anyway, so really it's just a half murder".
The arguments for abortion rights usually revolve around the body belonging to the woman. Well, her body certainly doesn't belong to some 3rd party murderous psychopath.
If you drive, you concentrate on the road, and if you don't do that, you face the consequences when the inevitable happens.
I agree, but consider the implications.
We have a speed limit law because people felt that there should be some objective criteria for determining reckless behavior before an accident.
Your attitude is: "Do whatever you want until some harm is done, then we sort it out in court."
I think that's a reasonable position, but one problem is that often the harm can't be undone by any court. You can't deter someone from speeding if they die in the accident anyway (since that's pretty much the maximum deterrent already). You can't bring the other person back to life.
So, as with many laws there is a compromise. If an officer is given objective criteria to deter, he can deter you from the (supposedly) dangerous practice by citing you. The practice may or may not be dangerous in that particular case, but the officer doesn't really need to prove that, since the law lists the objective criteria itself as illegal.
Now, we run into conflict-of-interest problems when the fine is paid back to the same people who issue the ticket, but that's another issue.
Nothing. I think a name means a lot, and Linus is able to protect that name and the products associated with Linux to an extent.
And yeah, it's just the kernel. I thought that's what this thread was talking about: a linux fork.
But Linus has a trademark on the term "Linux". He could probably prevent you from calling your forked OS "Linux" is he wanted to.
Let's assume your comment is correct. Why do we listen to any countries on the U.N.? Heck, LIBYA is on the human rights committee.
So these people who say we should be "multilateral" are really saying that we should follow the advice of a bunch of dictators who don't let their people speak. "Multilateral" sounds a lot better until you start listing the actual countries out, huh?
I suppose we should follow those dictators, because if we ignored them we'd be imposing democracy on other people.
That's why the U.N. is a joke. It's a joke when they don't enforce their resolutions. It's a joke when Libya is on the human rights committee. It's a joke when we pretend that a dictator speaks for his people.
If Spain wants to join the cause, then encourage terrorism by giving the terrorists results ("Hey, let's make some more demands in Spain, they seem to buckle quickly"), that's their problem. They're a sovereign nation, after all. At least until the next group of tyrants occupies their country and the U.S. needs to bail 'em out again.
Regardless, the U.K. is hardly anything less than 1st world, and they are a strong U.S. ally.
Probably the best example of irony that I heard was: "It's ironic if you poke yourself in the eye putting on a pair of safety glasses".
When someone says that something is ironic, I always compare it to that. This thing about a slashdot editor supposedly being annoying and posting a story about annoying people is really nothing worth HAVING a word for. You could call him a hypocrite, or dense or something, but really nobody cares and I don't think a new word is necessary.
And England and Spain...
never mind that most of the taxes are paid by the rich anyways.
The more important question is so much simpler. Are taxes on the rich to low? To high? Just right? Most people who whine about "tax cuts for the rich" have no idea what taxes on the "rich" actually are, or who is considered "rich". If "tax cuts for the rich" are always bad, then we should just tax "rich" people at 100%. So, sometimes a tax cut is good, sometimes it's bad. It all depends on the numbers, and the associated economic implications (which affect everyone).
Few people have enough time to bother themselves with numbers, facts, or real economic analysis. It's so much easier to just buy into some rhetoric.
Of course, there could be legitimate concerns about said tax cut. I'm just critisizing the people that are taken in by the rhetoric without asking the obvious questions.
Let's say a program is licensed with BSD. The theory behind the GPL says that a corporation can make a closed version of the same software, but add in a bunch of "must-have" features and lure people to use their improved (but closed) version. Short-sighted people might be lured away and the open version would disappear.
But what are the realities? MySQL and PHP both used to be more free than they are now. MySQL used to have LGPL client libraries and GPL code, now the client libraries are GPL (which is so restrictive that many people need to buy commercial licenses from MySQL when all they want to do is add a MySQL DB driver to their application). PHP used to be GPL, now it's not.
I guess the moral of the story is that the most important thing is that the copyright is not held by any one entity. You can debate whether GPL is better than BSD, but it seems a moot point when one entity controls the entire copyright, because they can do tricky things by changing the license themselves (like mysql client libraries going LGPL -> GPL).
In that respect something like PostgreSQL is much safer and more likely to remain free because the copyright is held by so many people that nobody could change the license.