I assume you're referring to Kelo v. New London. The interesting thing about that case is that all the dissenters were the conservatives. O'Connor, Rhenquist, Scalia, and Thomas all dissented from that horrible opinion.
I re-read your post and it still seems to have a strong negative attitude about relational database systems. It seems to suggest that they are somehow a legacy that has been outdated.
Show me some concrete data that prove RDMBS are the "best" thing out there as you claim.
Best is subjective. But, despite a huge amount of investment, "research", and marketing, neither OODBMSs nor XML Database Systems have made much progress into useful production applications. I would claim that's because those are a dead end and will never be useful enough to overcome the current relational database market.
I guarantee in many situations both in terms of productivity for developers and database performance that RDBMS get smoked.
By "productivity for developers" do you mean "got the first iteration of some application finished" or "created something that will usefully collect and manage data over a long period of time"? I suspect that OODBMs may be faster to accomplish the former, but I doubt they are more effective at the latter.
There's no denying things like ORMs and the mess the relational world has created.
That the relational world created? I think ORMs are something the OO world created. Interesting opinion, however.
If your only point is that OODBMSs are better than ORMs, you may be able to convince me of that. ORMs are certainly a mess, I'll give you that.
It seems like no one knows that object databases have been around a long time too.
Actually longer. An "object database" is really just a graph database. Objects refer to eachother via pointer chains, and it forms a graph. There's really no extra semantic value there, object databases are just graph databases adapted to OO development styles.
Relational database systems were invented to overcome the semantic problems with graph databases, not the performance problems.
One answer why object databases are ignored...
Here you start a list of very condescending remarks about relational database systems. Have you considered the possibility that many people use relational database systems because they are informed and implementing a new application; not in spite of those facts?
...to continue to hang on to our old RDBMSs
More of the same. Object databases aren't newer or better, they are repackaged graph databases. Businesses are "hanging on" to RDBMSs because they are the best thing out there (although you may be able to criticize some specific implementation).
I think you should really consider whether you're looking for a DBMS at all. If all you want is persistence, you should consider solutions along those lines.
Everybody is against [hands full cellphone calls while driving]
Really? What makes that worse than hands-free cellphone calls?
In fact, the linked study seems to suggest that the distracting part is the speaking/listening, not what you're holding.
This absolutism has no place in policy making. There are some very reasonable situations where making a phone call (hands-free or not) while driving is perfectly acceptable. For instance, if there is nobody else on the road, and you have perfect visibility around you, and you are moving at a speed safe for all of the conditions.
There are so many other things that matter so much more that are perfectly legal. What about driving without adequate sleep?
When on the road there is only one thing that is important and that is safety.
People who speak in absolutes are always ignoring the costs. There are obviously things that are important on the road in addition to safety, and sometimes in conflict with safety.
For instance, getting where you are going in a reasonable amount of time is important. Otherwise, we could strictly enforce a 5mph speed limit. That would eliminate nearly all fatal accidents. However, we are willing, as people in the real world, to allow thousands of people to die on the road as a result of allowing potentially fatal speeds on the road (like 30mph).
Talking on cell phones may be important to some people who prefer to be in touch with other people while driving. Prohibition of this activity has some cost.
Now, to take a perfectly reasonable scenario, let's say someone is in traffic and unable to pull over. They may be completely stopped or they may be moving at 5mph. Nobody will die as a result of them speaking on the phone, even if they do cause an accident. So someone might reasonably conclude that the person making the call should be the one to weigh the costs (which probably don't exist at all in 5mph traffic) versus the benefits.
Reckless endangerment is a legal tool to make otherwise lawful actions illegal because the risk is unacceptable -- even if nothing bad happens in that particular situation. But this is being extended to ridiculous extremes. Talking on a cell phone while driving is not reckless.
Given these facts, you understand that there are risks to driving, and you are free to choose not to drive. A bus is virtually immune from a mistake made by a small car, perhaps that's more suited to your tastes?
But there are reasons that such an attempt at a fork will ultimately fail. All of the developers are employed by Sun, and won't participate in the fork. So that means you have to build a developer community from nothing, which is almost impossible unless there's a lot of motivation. With an alternative like PostgreSQL, nobody's going to be motivated enough to learn the MySQL source code.
People throw around words like "fork" as if it were a realistic alternative for MySQL, but it's not.
If there were a big divide in the PostgreSQL community, chances are there would be a lot of developers on both sides, because they don't all work for the same company.
There are fundamental organizational differences between the two projects that matter much more than which licenses they happen to use.
That's pretty much what Codd and Date were using as a definition.
I don't know Codd's work well enough to disagree. But Chris Date and Hugh Darwen would almost certainly disagree strongly with your perception of the relational model.
Quacks like a duck, might as well be a duck.
The SQL standard is already imperfect, and MySQL deviates further from the relational model in important ways.
Let me guess, you guys are Oracle fans, no?
No. Although as an RDBMS, it's a respectable second-best to PostgreSQL.
MySQL in particular allows use of multiple storage managers.
But to be called a relational database management system, choosing a different storage engine should not affect the behavior, only performance.
In MySQL it does affect behavior to use a different storage engine, so it fails to abstract the relations from the physical storage.
I suppose you could argue that MySQL is really a lot of different relational database management systems combined in one package, but it still fails to enforce domain constraints, so it still fails to be an RDBMS.
Focuses on how the data can be accessed by and is presented to the using process.
That changes radically depending on the storage engine and various configuration options.
If you can access it with standard SQL...
MySQL does not enforce the SQL standard in many fundamental ways, unless I'm seriously mistaken about the SQL standard (which I may be). Does the SQL standard really allow such disregard for type constraints and foreign keys? Are backticks really an acceptable way to quote identifiers according to the standard?
there appears to be nothing that prevenst a PG developer form doing exactly what Sun is
From a legal standpoint, maybe. Apparently the GPL isn't really helping keep MySQL free either, because the MySQL code is owned entirely by Sun, so they can do whatever they want. For instance, if Sun develops a crucial feature and makes it closed, like online backup, they can prevent the GPL versions from being accepted to the project, and can let alternatives to that closed solution fall into unmaintained disrepair.
In contrast, the PostgreSQL community is a group of many individuals and companies, and they cannot prevent free versions of important features from being committed.
So really, the question is not what could be done legally, but what actually is happening, and what is likely to continue happening. MySQL started as a commercial product, went GPL, and now they are introducing closed features again. They always maintained copyright ownership.
Again, in contrast, the PostgreSQL community started with a free license, and has a wide group of contributors that each retain copyright for their respective contributions. Perhaps most importantly, PostgreSQL has very well-commented, readable source code.
RHEL backpatches a lot of stability and security fixes from newer software. This means that you have the benefit of the new fixes, without the drawback of the new flaws.
I'm impressed by your faith in the GPL. All of the coders work for Sun now, so that means everyone who started up the fork would be new to the codebase.
I haven't looked at that code, but it doesn't have a reputation for being clean, readable, well-commented code -- especially compared to PostgreSQL.
I have my doubts that the project could pick up where Sun left off in any reasonable amount of time. If there were no other options, it would be one thing. But PostgreSQL is a great option, so why would people spend so much effort trying to rebuild the MySQL developer base from scratch?
I mostly agree, but I'd like to add that there are all of the application developers that have made MySQL so popular. Sun needs those people, and if they jump ship (to a fork, or to PostgreSQL), that will be a problem for MySQL.
The other problem with Postgres is that there is literally no way to implement a HA Cluster with any kind of reliability.
It depends on what you mean by "HA". If you can stand a couple seconds of downtime, and loss of client connections, you can use a "warm standby" system. And that is reliable, unless you have other information.
You don't want to rsync a data directory on a running server. Either shut the server down, or use a filesystem-level snapshot.
"Mirroring" isn't as easy to do as it is to say, unfortunately. Depending on exactly how you're using it, there are a lot of pitfalls. This is true for any replication system, not just postgres.
Having recently seen Sun buy MySQL, this looks a lot like a "me too"-move.
"Astor [Andy Astor, CEO of EnterpriseDB] added that the discussions with IBM pre-dated the $1 billion acquisition of MySQL by Sun (NASDAQ: JAVA) earlier this year."
How does associating with someone who has said something like that make Obama unfit to be president?
I didn't say it did. But it's certainly not good, particularly when his wife is saying things like "for the first time in my life I'm proud of my country" (paraphrased).
If a white person was attending sermons with that level of hatred for 20 years, and donated $20k, that person's candidacy would be tanked. Obama is now relying on that double standard. And the whole time Obama was supposed to be above the race issue, which was why I originally respected him.
At absolute minimum, Obama had to know that associating so closely with Wright lent credibility to his hateful speech.
I've issues with that policy, but I'm failing to see how it's not equal protection under the law - they didn't apply the policy in a discriminatory way. It's a policy applied equally to all 50 states, last time I checked.
By this logic, there are no limits to the power of the federal government to coerce the states financially. They simply take away the money, and give it back conditionally, where the conditions are whatever the federal government wants.
It discriminates against states that are exercising their free will on issues that the federal government should have no part in. The federal government's laws should not make reference to those powers reserved for the states.
It isn't, and I'd be interested to see your evidence that McCain, Obama, and Clinton spoke in favour of that decision.
All of them take "commerce among the states" to mean virtually any activity at all. Any candidate who favors federal regulation of guns, crops, or drugs that do not cross state lines is implicitly relying on decisions like this one.
As with the first quote, I'd like to see the evidence that Paul is the only one opposed to this.
I never said he was.
And they have both the power to tax and the power to spend.
They only have the power to spend in order to exercise the powers that they do have. The Constitution does not allow them to spend however they see fit.
That's right, we have a republic, based in part on democracy. An entire branch of government is entirely undemocratic.
How many democratic countries do you know where "senior" and "junior" ruled? If everybody really had equal chances the possibility is near zero.
Democracy doesn't mean "everyone has equal chance to be a leader", it means "the popular vote has the power" (which is partially true in the US). There is no reason to think that popularity would be spread evenly among the people; quite the contrary.
You guys got a moneycracy.
If that were true, John Kerry would have won the 2004 election, or Edwards. There are many examples of someone with less money winning. I think there is a reasonable argument that you need some baseline of money to run for high office, but it certainly is not a matter of "the most money wins".
Slashdot Mirror
I assume you're referring to Kelo v. New London. The interesting thing about that case is that all the dissenters were the conservatives. O'Connor, Rhenquist, Scalia, and Thomas all dissented from that horrible opinion.
I re-read your post and it still seems to have a strong negative attitude about relational database systems. It seems to suggest that they are somehow a legacy that has been outdated.
Show me some concrete data that prove RDMBS are the "best" thing out there as you claim.
Best is subjective. But, despite a huge amount of investment, "research", and marketing, neither OODBMSs nor XML Database Systems have made much progress into useful production applications. I would claim that's because those are a dead end and will never be useful enough to overcome the current relational database market.
I guarantee in many situations both in terms of productivity for developers and database performance that RDBMS get smoked.
By "productivity for developers" do you mean "got the first iteration of some application finished" or "created something that will usefully collect and manage data over a long period of time"? I suspect that OODBMs may be faster to accomplish the former, but I doubt they are more effective at the latter.
There's no denying things like ORMs and the mess the relational world has created.
That the relational world created? I think ORMs are something the OO world created. Interesting opinion, however.
If your only point is that OODBMSs are better than ORMs, you may be able to convince me of that. ORMs are certainly a mess, I'll give you that.
It seems like no one knows that object databases have been around a long time too.
...to continue to hang on to our old RDBMSs
Actually longer. An "object database" is really just a graph database. Objects refer to eachother via pointer chains, and it forms a graph. There's really no extra semantic value there, object databases are just graph databases adapted to OO development styles.
Relational database systems were invented to overcome the semantic problems with graph databases, not the performance problems.
One answer why object databases are ignored...
Here you start a list of very condescending remarks about relational database systems. Have you considered the possibility that many people use relational database systems because they are informed and implementing a new application; not in spite of those facts?
More of the same. Object databases aren't newer or better, they are repackaged graph databases. Businesses are "hanging on" to RDBMSs because they are the best thing out there (although you may be able to criticize some specific implementation).
I think you should really consider whether you're looking for a DBMS at all. If all you want is persistence, you should consider solutions along those lines.
How about in addition to being distracted, you also have little chance to fully control the car when something happens.
Do you have any evidence to support your hypothesis?
What about manual transmissions, should we outlaw those? Shifting requires that your ten-and-two rule be broken.
Everybody is against [hands full cellphone calls while driving]
Really? What makes that worse than hands-free cellphone calls?
In fact, the linked study seems to suggest that the distracting part is the speaking/listening, not what you're holding.
This absolutism has no place in policy making. There are some very reasonable situations where making a phone call (hands-free or not) while driving is perfectly acceptable. For instance, if there is nobody else on the road, and you have perfect visibility around you, and you are moving at a speed safe for all of the conditions.
There are so many other things that matter so much more that are perfectly legal. What about driving without adequate sleep?
When on the road there is only one thing that is important and that is safety.
People who speak in absolutes are always ignoring the costs. There are obviously things that are important on the road in addition to safety, and sometimes in conflict with safety.
For instance, getting where you are going in a reasonable amount of time is important. Otherwise, we could strictly enforce a 5mph speed limit. That would eliminate nearly all fatal accidents. However, we are willing, as people in the real world, to allow thousands of people to die on the road as a result of allowing potentially fatal speeds on the road (like 30mph).
Talking on cell phones may be important to some people who prefer to be in touch with other people while driving. Prohibition of this activity has some cost.
Now, to take a perfectly reasonable scenario, let's say someone is in traffic and unable to pull over. They may be completely stopped or they may be moving at 5mph. Nobody will die as a result of them speaking on the phone, even if they do cause an accident. So someone might reasonably conclude that the person making the call should be the one to weigh the costs (which probably don't exist at all in 5mph traffic) versus the benefits.
Reckless endangerment is a legal tool to make otherwise lawful actions illegal because the risk is unacceptable -- even if nothing bad happens in that particular situation. But this is being extended to ridiculous extremes. Talking on a cell phone while driving is not reckless.
Given these facts, you understand that there are risks to driving, and you are free to choose not to drive. A bus is virtually immune from a mistake made by a small car, perhaps that's more suited to your tastes?
Keep debugging and rewriting until you stumble upon a working implementation?
It works great for user interfaces. It's terrible for systems with long term consequences.
Unfortunately, there are many people who can't distinguish between the two.
Further, there are huge advantages to having 100% control over the memory management of your application
malloc()/free() are memory managers as well. So I suppose you must use brk()/sbrk() instead, right?
There is no reason the code can't be forked;
But there are reasons that such an attempt at a fork will ultimately fail. All of the developers are employed by Sun, and won't participate in the fork. So that means you have to build a developer community from nothing, which is almost impossible unless there's a lot of motivation. With an alternative like PostgreSQL, nobody's going to be motivated enough to learn the MySQL source code.
People throw around words like "fork" as if it were a realistic alternative for MySQL, but it's not.
If there were a big divide in the PostgreSQL community, chances are there would be a lot of developers on both sides, because they don't all work for the same company.
There are fundamental organizational differences between the two projects that matter much more than which licenses they happen to use.
That's pretty much what Codd and Date were using as a definition.
I don't know Codd's work well enough to disagree. But Chris Date and Hugh Darwen would almost certainly disagree strongly with your perception of the relational model.
Quacks like a duck, might as well be a duck.
The SQL standard is already imperfect, and MySQL deviates further from the relational model in important ways.
Let me guess, you guys are Oracle fans, no?
No. Although as an RDBMS, it's a respectable second-best to PostgreSQL.
The benefits of sticking with a widely used standard
You mean like SQL? PostgreSQL supports it. MySQL doesn't. So I don't follow your argument here.
MySQL in particular allows use of multiple storage managers.
...
But to be called a relational database management system, choosing a different storage engine should not affect the behavior, only performance.
In MySQL it does affect behavior to use a different storage engine, so it fails to abstract the relations from the physical storage.
I suppose you could argue that MySQL is really a lot of different relational database management systems combined in one package, but it still fails to enforce domain constraints, so it still fails to be an RDBMS.
Focuses on how the data can be accessed by and is presented to the using process.
That changes radically depending on the storage engine and various configuration options.
If you can access it with standard SQL
MySQL does not enforce the SQL standard in many fundamental ways, unless I'm seriously mistaken about the SQL standard (which I may be). Does the SQL standard really allow such disregard for type constraints and foreign keys? Are backticks really an acceptable way to quote identifiers according to the standard?
there appears to be nothing that prevenst a PG developer form doing exactly what Sun is
From a legal standpoint, maybe. Apparently the GPL isn't really helping keep MySQL free either, because the MySQL code is owned entirely by Sun, so they can do whatever they want. For instance, if Sun develops a crucial feature and makes it closed, like online backup, they can prevent the GPL versions from being accepted to the project, and can let alternatives to that closed solution fall into unmaintained disrepair.
In contrast, the PostgreSQL community is a group of many individuals and companies, and they cannot prevent free versions of important features from being committed.
So really, the question is not what could be done legally, but what actually is happening, and what is likely to continue happening. MySQL started as a commercial product, went GPL, and now they are introducing closed features again. They always maintained copyright ownership.
Again, in contrast, the PostgreSQL community started with a free license, and has a wide group of contributors that each retain copyright for their respective contributions. Perhaps most importantly, PostgreSQL has very well-commented, readable source code.
RHEL backpatches a lot of stability and security fixes from newer software. This means that you have the benefit of the new fixes, without the drawback of the new flaws.
It also avoids breaking backward compatibility.
I'm impressed by your faith in the GPL. All of the coders work for Sun now, so that means everyone who started up the fork would be new to the codebase.
I haven't looked at that code, but it doesn't have a reputation for being clean, readable, well-commented code -- especially compared to PostgreSQL.
I have my doubts that the project could pick up where Sun left off in any reasonable amount of time. If there were no other options, it would be one thing. But PostgreSQL is a great option, so why would people spend so much effort trying to rebuild the MySQL developer base from scratch?
There are close to no free coders involved.
I mostly agree, but I'd like to add that there are all of the application developers that have made MySQL so popular. Sun needs those people, and if they jump ship (to a fork, or to PostgreSQL), that will be a problem for MySQL.
The other problem with Postgres is that there is literally no way to implement a HA Cluster with any kind of reliability.
It depends on what you mean by "HA". If you can stand a couple seconds of downtime, and loss of client connections, you can use a "warm standby" system. And that is reliable, unless you have other information.
Postgres 8.4 will have Postgres-R built-in
What makes you say that?
You don't want to rsync a data directory on a running server. Either shut the server down, or use a filesystem-level snapshot.
"Mirroring" isn't as easy to do as it is to say, unfortunately. Depending on exactly how you're using it, there are a lot of pitfalls. This is true for any replication system, not just postgres.
Having recently seen Sun buy MySQL, this looks a lot like a "me too"-move.
"Astor [Andy Astor, CEO of EnterpriseDB] added that the discussions with IBM pre-dated the $1 billion acquisition of MySQL by Sun (NASDAQ: JAVA) earlier this year."
-- http://www.internetnews.com/software/article.php/3736336
They have the power to provide for the general welfare, remember?
That's only in the preamble, and not a power granted.
How does associating with someone who has said something like that make Obama unfit to be president?
I didn't say it did. But it's certainly not good, particularly when his wife is saying things like "for the first time in my life I'm proud of my country" (paraphrased).
If a white person was attending sermons with that level of hatred for 20 years, and donated $20k, that person's candidacy would be tanked. Obama is now relying on that double standard. And the whole time Obama was supposed to be above the race issue, which was why I originally respected him.
At absolute minimum, Obama had to know that associating so closely with Wright lent credibility to his hateful speech.
I still think he'd be better than Hillary.
I've issues with that policy, but I'm failing to see how it's not equal protection under the law - they didn't apply the policy in a discriminatory way. It's a policy applied equally to all 50 states, last time I checked.
By this logic, there are no limits to the power of the federal government to coerce the states financially. They simply take away the money, and give it back conditionally, where the conditions are whatever the federal government wants.
It discriminates against states that are exercising their free will on issues that the federal government should have no part in. The federal government's laws should not make reference to those powers reserved for the states.
It isn't, and I'd be interested to see your evidence that McCain, Obama, and Clinton spoke in favour of that decision.
All of them take "commerce among the states" to mean virtually any activity at all. Any candidate who favors federal regulation of guns, crops, or drugs that do not cross state lines is implicitly relying on decisions like this one.
As with the first quote, I'd like to see the evidence that Paul is the only one opposed to this.
I never said he was.
And they have both the power to tax and the power to spend.
They only have the power to spend in order to exercise the powers that they do have. The Constitution does not allow them to spend however they see fit.
There are non-economic costs to socialism as well.
But that's beside the point, because you missed the post I was responding to:
Grandparent: McCain might help the economy...
Parent: I really wouldn't count on Mr. "Hundred Years in Iraq" to do that...
So, economic costs are the only costs at issue, in this thread.
But you havent got a democracy.
That's right, we have a republic, based in part on democracy. An entire branch of government is entirely undemocratic.
How many democratic countries do you know where "senior" and "junior" ruled? If everybody really had equal chances the possibility is near zero.
Democracy doesn't mean "everyone has equal chance to be a leader", it means "the popular vote has the power" (which is partially true in the US). There is no reason to think that popularity would be spread evenly among the people; quite the contrary.
You guys got a moneycracy.
If that were true, John Kerry would have won the 2004 election, or Edwards. There are many examples of someone with less money winning. I think there is a reasonable argument that you need some baseline of money to run for high office, but it certainly is not a matter of "the most money wins".