It think the problem with saudi arabia is that the money is coming from top to the bottom. This allows for a very dangerous situation where wealthy patrons can finance/coerce those with less money to do their will. Without an independent source of supporting themselves, the young turn to the ones who say they are from God and offering them and their families money and protection. It was only really when the masses were able to provide for themselves that they were able to make more responsible decisions for themselves and their families.
Yeah, I think I understand what you are saying. Its not as trivial as it might first seem. Mysql doesn't have sequences, (which allows it to ignore that problem), but that means it uses auto-incrementing columns which are less flexible and require a pseudo table lock on inserts into Innodb tables which limits its concurrency (impact lessened with 5.1 and row based replication). Believe me, I wish I had access to Postgres style sequences.
If I could edit that, I would. I didn't mean to be so dismissive of your tuna related ignorance. If you take offence, I apologize. While at the same time, you might as well consider broadening your horizons as well. Bluefin is so widely demanded, that its on the verge of being fished to extinction. I assure you, none of it ends up in a can.
Where do you come from? I'm pretty sure based on your answer its not Japan, Italy, New York, LA, Hawaii, Seattle, or Miami. Tuna is delicious ( when not served out of a can). Mayo is *not* welcome on my Tuna steaks or sushi.
Ok, thanks for the explanation. That makes more sense. Traditionally, Mysql uses Statement based replication. So the whole statement was relayed to the slave for it to execute. This allows it to avoid this problem.
Row based replication is fairly new in MySql, and sounds closer to what you are describing. I'll need to research the details of that, but I'm 80% sure it doesn't have this problem with replication conflicts.
Personally, I love statement based replication. I'm also aware that Postgres has some really nice features and better Vertical scalability. I would love to benchmark them against each other, but this replication behaviour pretty much rules that out. Its not a viable solution for long running queries on a slave. We need to be able to run long running queries on slaves that are not far behind in replication. MyISam cannot do MVCC, and thus would block replication during the long running query. The servers would be days behind in replication. This was a pretty bad situation for us, so we switched to Innodb for just this reason.
The other issue is that the VACUUM process removes rows from the database that are no longer visible to any transactions it knows about, and this too is written in to the WAL and communicated to the standby server, so VACUUMing the primary can delete rows that old transactions on the standby server are still looking at.
Why doesn't that just work? With Mysql's INNODB storage engine, by default each transaction is isolated from each other (known as the READ_REPEATABLE isolation level). If you start a transaction in Connection A, and only do reads from the database it will continue to see all of the records it has examined. They won't change. So if on Connection B, a transaction has Begun after the the transaction in Connection A and Deleted a row that transaction B selected from, there is no conflict. Every thing is fine. Its know as READ_REPEATABLE isolation level. It can be configured to an isolation level similar to what you are describing, known as READ_COMMITTED, but most people don't do that for the very same reason you are mentioning. Can't you get PostGres to operate with a similar level of isolation?
Yes there were third party tools each with a different foucs, none of which was as adaptable as Mysql's built in replication support.
Can you explain the new built in replication's conflict handling to me? I don't understand why the conflicts would happen with a multi-version concurrency control. According to them, a Delete coming in from the master would conflict with a long running read only query on the slave. That really should not happen if you have true MVCC, and is a real deal breaker.
As the other responder noted, thats why one would promote a slave in that scenario.
ereg sucks. Preg sucks less. Choosing the one with worse performance and less features, and never examining your decision is not smart. I'm in no way defending the idiot PHP developers, rather I'm blaming Idiot PHP Developers that made a horrible choice because of idiotic decisions the PHP developers made. IN any case as I've said before and will continue to say: languages evolve. developers that don't will become roadkill. You can't complain when a shitty piece of a language is removed, especially when there has almost always been a better choice available. Stop eating dirt, have you heard about chicken? It tastes better and has less rocks in in it.
I'm with the other commenter's, ereg is just as bad as magic_quotes, IMHO. Its like you went to a Yugo/Honda dealer saw that the Honda Civic was priced lower than the Yugo, and still bought the Yugo because you thought Yugo wouldn't change its product line as much.
Yeah, you really do have to keep up with a language these days to keep an eye out for the future. I wonder how you managed the transition through the other PHP versions. I mean, you aren't using register_globals or magic_quotes are you? Languages evolve. Programmers must evolve with them. If you don't then you'll get left behind.
I don't see how going to a non-dead tree platform means they can't come out with new editions. Although, with enough DRM they can force all students to buy *new* copies of what every edition they'd like without having to worry about the used book market.
What size (MB and #lines of host files) of a dataset did you actually have?
Why did you use Delphi 7, exactly? Its just an odd choice (from 2001), not universally proclaimed as the fastest in the world. Is it because its the fastest that you are familiar with? Its also obviously not 64 bit.
Why are you doing server work on a desktop? Typically that kind of a workload is performed on servers. It doesn't really require a gui, just brunt horsepower. It isn't something that most desktop users would be doing, not even power users like graphic designers or cad users. There is no argument that you could benefit from faster processors in a server. However, that is not what we are talking about here.
Why are you comparing the x2 4800 from 2005 to a top of the line processor from today? That's not a $100 processor of today, its a discontinued not even available in most web stores, processor. $100 gets you a amd phenom II 3.2 ghz processor today. See how much of a difference there is between that and your i7.
Way back when, before you remember, firefox was supposed to be integrated into the next version of mozilla called "Seamonkey". Seamokey was the true inheritor of the netscape communicator internet suite line. Why would any internet user not want a consistent interface for web browsing, emailing, website creation, and other stupid things no one else does? But, before that could happen firefox caught fire. People loved it, despite the lack of a html editor. So they cancelled the seamonkey plans, and threw those same developers at firefox to help "improve" it. So you had a bunch of programmers used to trying to build a fully featured internet suite, now working on "just a browser". Well, this is what you got firefox, the bloated. All the features you could want in a browser, memory, code bloat be danged.
Sadistically, some developers rebelled against the "light weight" of firefox and actually went a head and built sea monkey as planned. They are truly disturbed individuals, who should be considered dangerous to themselves and other people. Please do not approach them yourself, but contact the local institute for the criminally insane.
And much like the typical consumer, I would respond to that: I don't care about that crap, 99% of it is the same. They are the same trucks. The only important difference is the gas mileage vs hauling capacity, the rest is all grill.
That's a good counter argument, but nothing more. In summary: Apple is #1, android is gaining. Apple recognises it has competition, its negative statements about the android market reflect that. If you want to characterise that as quaking or not is a semantic argument best left to marketing types.
Now, what will the future hold? I don't know. You don't know. Apple and Google doesn't know. If you tried arguing with some one back in 1991 that Apple was going to end up on the brink of bankruptcy in six years and would have to rely on Microsoft to prop them up ( due to MS's dominance based on a crappy, buggy operating system that crashed more often than it worked) you'd have called me crazy. Leave the far off predictions to the insane I say. Choose what seems best for you over the short term, with disposable products like cell phones.
Well, Holland is actually in Michigan on the east coast of Lake Michigan. They make some pretty good beer there too. Plus wooden shoes, tulips, and everything else Dutch.
When you hear the word "Sexting" keep in mind that its minors ( those under the age of consent) sending pictures of themselves. No one otherwise cares about consenting adults doing it. The Kin, as microsoft has indicated is targeted towards teenagers. Those same teenagers that sext. Do I need to draw you a map?
No, not joking. The add promotes taking pictures of oneself under clothing and sending them to others.Its not about the functionality, its the promotion of the idea that taking pictures of oneself under clothing and sending them to others is an acceptable thing that is easily accomplished with this device. Sure there are others that have that capability ( hence texting's rampant popularity), but the phone makers do not advertise that capability nor promote it. Everyone here seems to be jumping on the non-sexuality of the male breast, rather than the slightly larger message ( you can take pictures of body parts and send them).
If one is serious about the problem of texting and does not want it promoted by phone makers in advertisements, then one has to draw a line where some picture taking in ads is okay and others is not. Two reasonable adults can disagree about where that line is, and I don't necessarily agree that this ad has crossed the line. But, I would agree that its close, and I understand the reasoning behind those that object to it. Few other slashdotters seem to understand why they are upset.
Slashdot Mirror
What and what now? Its like you're making up words.
It think the problem with saudi arabia is that the money is coming from top to the bottom. This allows for a very dangerous situation where wealthy patrons can finance/coerce those with less money to do their will. Without an independent source of supporting themselves, the young turn to the ones who say they are from God and offering them and their families money and protection. It was only really when the masses were able to provide for themselves that they were able to make more responsible decisions for themselves and their families.
Except there are five, and I suck at reading comprehension. I must be a regular here.
No three persons, one substance. Sounds like the Trinity to me. Christians win.
Wonderful! I'll take 12!
I know what the implied answer to that question is, I'm not sure how much I believe it. Reason #213 why I am not using MS software any more.
Yeah, I think I understand what you are saying. Its not as trivial as it might first seem. Mysql doesn't have sequences, (which allows it to ignore that problem), but that means it uses auto-incrementing columns which are less flexible and require a pseudo table lock on inserts into Innodb tables which limits its concurrency (impact lessened with 5.1 and row based replication). Believe me, I wish I had access to Postgres style sequences.
If I could edit that, I would. I didn't mean to be so dismissive of your tuna related ignorance. If you take offence, I apologize. While at the same time, you might as well consider broadening your horizons as well. Bluefin is so widely demanded, that its on the verge of being fished to extinction. I assure you, none of it ends up in a can.
Where do you come from? I'm pretty sure based on your answer its not Japan, Italy, New York, LA, Hawaii, Seattle, or Miami. Tuna is delicious ( when not served out of a can). Mayo is *not* welcome on my Tuna steaks or sushi.
Ok, thanks for the explanation. That makes more sense. Traditionally, Mysql uses Statement based replication. So the whole statement was relayed to the slave for it to execute. This allows it to avoid this problem.
Row based replication is fairly new in MySql, and sounds closer to what you are describing. I'll need to research the details of that, but I'm 80% sure it doesn't have this problem with replication conflicts.
Personally, I love statement based replication. I'm also aware that Postgres has some really nice features and better Vertical scalability. I would love to benchmark them against each other, but this replication behaviour pretty much rules that out. Its not a viable solution for long running queries on a slave. We need to be able to run long running queries on slaves that are not far behind in replication. MyISam cannot do MVCC, and thus would block replication during the long running query. The servers would be days behind in replication. This was a pretty bad situation for us, so we switched to Innodb for just this reason.
So, if I set everything up correctly, would I be able to get similar behaviour from PostGres for Replication? IE, no conflicts?
Why doesn't that just work? With Mysql's INNODB storage engine, by default each transaction is isolated from each other (known as the READ_REPEATABLE isolation level). If you start a transaction in Connection A, and only do reads from the database it will continue to see all of the records it has examined. They won't change. So if on Connection B, a transaction has Begun after the the transaction in Connection A and Deleted a row that transaction B selected from, there is no conflict. Every thing is fine. Its know as READ_REPEATABLE isolation level. It can be configured to an isolation level similar to what you are describing, known as READ_COMMITTED, but most people don't do that for the very same reason you are mentioning. Can't you get PostGres to operate with a similar level of isolation?
Yes there were third party tools each with a different foucs, none of which was as adaptable as Mysql's built in replication support.
Can you explain the new built in replication's conflict handling to me? I don't understand why the conflicts would happen with a multi-version concurrency control. According to them, a Delete coming in from the master would conflict with a long running read only query on the slave. That really should not happen if you have true MVCC, and is a real deal breaker.
As the other responder noted, thats why one would promote a slave in that scenario.
ereg sucks. Preg sucks less. Choosing the one with worse performance and less features, and never examining your decision is not smart. I'm in no way defending the idiot PHP developers, rather I'm blaming Idiot PHP Developers that made a horrible choice because of idiotic decisions the PHP developers made. IN any case as I've said before and will continue to say: languages evolve. developers that don't will become roadkill. You can't complain when a shitty piece of a language is removed, especially when there has almost always been a better choice available. Stop eating dirt, have you heard about chicken? It tastes better and has less rocks in in it.
I'm with the other commenter's, ereg is just as bad as magic_quotes, IMHO. Its like you went to a Yugo/Honda dealer saw that the Honda Civic was priced lower than the Yugo, and still bought the Yugo because you thought Yugo wouldn't change its product line as much.
Yeah, you really do have to keep up with a language these days to keep an eye out for the future. I wonder how you managed the transition through the other PHP versions. I mean, you aren't using register_globals or magic_quotes are you? Languages evolve. Programmers must evolve with them. If you don't then you'll get left behind.
I don't see how going to a non-dead tree platform means they can't come out with new editions. Although, with enough DRM they can force all students to buy *new* copies of what every edition they'd like without having to worry about the used book market.
Interesting post.
Few questions unanswered by it:
What size (MB and #lines of host files) of a dataset did you actually have?
Why did you use Delphi 7, exactly? Its just an odd choice (from 2001), not universally proclaimed as the fastest in the world. Is it because its the fastest that you are familiar with? Its also obviously not 64 bit.
Why are you doing server work on a desktop? Typically that kind of a workload is performed on servers. It doesn't really require a gui, just brunt horsepower. It isn't something that most desktop users would be doing, not even power users like graphic designers or cad users. There is no argument that you could benefit from faster processors in a server. However, that is not what we are talking about here.
Why are you comparing the x2 4800 from 2005 to a top of the line processor from today? That's not a $100 processor of today, its a discontinued not even available in most web stores, processor. $100 gets you a amd phenom II 3.2 ghz processor today. See how much of a difference there is between that and your i7.
Way back when, before you remember, firefox was supposed to be integrated into the next version of mozilla called "Seamonkey". Seamokey was the true inheritor of the netscape communicator internet suite line. Why would any internet user not want a consistent interface for web browsing, emailing, website creation, and other stupid things no one else does? But, before that could happen firefox caught fire. People loved it, despite the lack of a html editor. So they cancelled the seamonkey plans, and threw those same developers at firefox to help "improve" it. So you had a bunch of programmers used to trying to build a fully featured internet suite, now working on "just a browser". Well, this is what you got firefox, the bloated. All the features you could want in a browser, memory, code bloat be danged.
Sadistically, some developers rebelled against the "light weight" of firefox and actually went a head and built sea monkey as planned. They are truly disturbed individuals, who should be considered dangerous to themselves and other people. Please do not approach them yourself, but contact the local institute for the criminally insane.
And much like the typical consumer, I would respond to that: I don't care about that crap, 99% of it is the same. They are the same trucks. The only important difference is the gas mileage vs hauling capacity, the rest is all grill.
That's a good counter argument, but nothing more. In summary: Apple is #1, android is gaining. Apple recognises it has competition, its negative statements about the android market reflect that. If you want to characterise that as quaking or not is a semantic argument best left to marketing types.
Now, what will the future hold? I don't know. You don't know. Apple and Google doesn't know. If you tried arguing with some one back in 1991 that Apple was going to end up on the brink of bankruptcy in six years and would have to rely on Microsoft to prop them up ( due to MS's dominance based on a crappy, buggy operating system that crashed more often than it worked) you'd have called me crazy. Leave the far off predictions to the insane I say. Choose what seems best for you over the short term, with disposable products like cell phones.
Well not Official Java, more like small pacific island coffee based beverage brand programming language for smart phones.
Wow that rolls off the tounge nicely. I should have gone into marketing generic products.
Well, Holland is actually in Michigan on the east coast of Lake Michigan. They make some pretty good beer there too. Plus wooden shoes, tulips, and everything else Dutch.
When you hear the word "Sexting" keep in mind that its minors ( those under the age of consent) sending pictures of themselves. No one otherwise cares about consenting adults doing it. The Kin, as microsoft has indicated is targeted towards teenagers. Those same teenagers that sext. Do I need to draw you a map?
No, not joking. The add promotes taking pictures of oneself under clothing and sending them to others.Its not about the functionality, its the promotion of the idea that taking pictures of oneself under clothing and sending them to others is an acceptable thing that is easily accomplished with this device. Sure there are others that have that capability ( hence texting's rampant popularity), but the phone makers do not advertise that capability nor promote it. Everyone here seems to be jumping on the non-sexuality of the male breast, rather than the slightly larger message ( you can take pictures of body parts and send them).
If one is serious about the problem of texting and does not want it promoted by phone makers in advertisements, then one has to draw a line where some picture taking in ads is okay and others is not. Two reasonable adults can disagree about where that line is, and I don't necessarily agree that this ad has crossed the line. But, I would agree that its close, and I understand the reasoning behind those that object to it. Few other slashdotters seem to understand why they are upset.