Ah, I see. I point out that it's a two-sided coin and call those who want it both ways wankers, and that's trolling.:P
It doesn't change the truth -- you can't have it both ways. Either security works and you have privacy/protection, or you can pirate to your heart's content. The writing is on the wall: society and legislation want the privacy.
The objectives have not changed, people are just beginning to understand that if I protect my information, no one else can access it. The reverse side of that equation is that if someone protects their information, I can't access it.
i.e. The technology that allows the *AA to come up with usage-restriction schemes is the same technology being mandated by privacy advocates when your address, financial, health care, and other information is accessed. It is expressly stated in the Canadian requirements for such systems that you must log every attempt to access a document.
So stop your freakin' panicking already, you anti-DRM wankers! Take it to the courts and fight the problem where it needs to be solved -- on a social/legal level.
That includes whoever came up with this flamebait phrasing:
Even worse: 'The system is also aimed at tracking who reads a document and when, because the chip can report back every access attempt. If you access the file, your fingerprint is recorded.'"
<sarcasm>
Yeah, but they don't wear brown shirts, red and black armbands, or thump the Koran this time. This time it's Americans thumping a puritan Bible, so it's ok, right?
</sarcasm>
No, the **AA does their best to force me to buy replacement media for as many playback device formats as I have access to.
The auto manufacturers, on the other hand, produce vehicles that last long enough to keep the customer satisfied if they put in the maintenance. They'd like you to buy more often, but they know that if their vehicles don't last, no one will buy them.
The **AA bully and intimidate everyone from barely-teens to aging grandmothers. They rely on lobbyists and bought politicians to "earn" their ever-increasing legal "rights" at the expense of what used to be consumer rights.
No business is guaranteed a market. If your market dries up and you don't find a way to adapt, you join the infamous buggy-whip maker. I'm sure they tried to push for legislation to make the "motor car" illegal, too.
The **AA has spent the past 20 years trying to change the rules. You used to be able to send back cassettes or albums for replacements when damaged; the only charge was shipping.
Now they tell you it can't be replaced, because that version has been replaced by a "new" release, even with relatively-recently purchased media.
Currently they're trying to cut it back further, so that it's not even legal for you to listen to your media on a portable device without paying yet again.
To hell with the greedy bastards. Once or twice at the trough was more than enough -- no more.
I actually buy a copy of Linux (usually SuSE) when I do an upgrade. Sure I could just download a couple.iso's and it's all legal, but I figure it's only fair to spend a few dollars when upgrading.
Buying a copy means some meagre portion of a developer's salary is paid to keep working on patches, updates, and integration of the overall system. For all the talk about open source, the truth is that commercial and OSS vendors pay for a lot of skilled developer resources to maintain the common source code.
Vendors provide support services, whether their code base is OSS or not.
A lot of industries also lend developer expertise, fixing bugs that affect the specific platform/tool combinations they're dealing with. Someone has to do it -- and a business needs it badly enough to pay to have it done.
Is it possible that this one-copy buyer sees OS/X as mostly OSS with a few GUI applications and yet another shiny desktop? How much is a desktop with utilities worth nowadays?
My understanding is that mainframe DB/2 lets you specify the type of storage and indexing to provide over data. PostgreSQL provides some control over that detail. MySQL was pretty much built as a virtual database over real containers. Sybase and Ingres have always allowed a fair bit of control over their indexing and container options. Oracle has some tuning options as well.
But if a vendor is targetting environments that need fine-tuning to eek every last bit of performance out of the hardware, it's always good to have more options for certain performance profiles. They're not common, but there might be a large enough customer base to consider adding that type of tuning to Oracle.
The fine-tuning options are all that significantly distinguish the top 3-4 RDBMS vendors. They all meet the checklists for multi-language, multi-platform drivers, SQL92+ standards support, integrity constraints, triggers, stored procs, replication options, online backup, and hot-failover and/or cluster support. You just need to buy into vendor-specific syntax to tweak the performance. It's getting hard to find specific areas where one vendor is "better" or "easier" than another.
The article may be a troll, but I notice no one is actually talking about network operating systems in response.
J2EE is a network OS programming kit. DotNet is another. QNX was designed as a networked/RPC version of *nix.
Given the market share the first two of that last have, I think the argument about whether network-bootable OS components/updates are viable is a moot point -- they've been here for years. Someone else just finally realized the fact.
The very first iteration of what eventually became Unix was a simple task switcher to allow a game to run at the same time as actual work. Technically it wasn't multi-user, because there was only a system console.
Actually I'm not talking about reject rules, I'm talking about confirmation-required flagging. So the only time an operator has to intervene is if flags are getting raised while processing a particular piece of data.
In some environments, there might even be multiple digital signatures required to approve such abnormal requests.
Some businesses just consider such checks and balances to be essential for fraud prevention and detection, but they can be used to catch a lot of other potentially expensive data problems early.
Not to defend outright piracy, but I know many people (sometimes myself as well) who rent and return movies within 24 hours. It only takes a couple hours to watch a movie, and it beats the heck out of TV tripe most of the time. Some of them will watch a couple movies a night.
So for them, this "heavy usage pattern" is perfectly normal and not a sign of piracy. Having their shipments "throttled" means they aren't getting the full use of the service they signed up for. It's up to the service provider to adjust capacity and/or pricing tiers to deal with the load, not choke off paid subscriptions that actually use the service as advertised.
Compared to the effort and expense of doing data range and argument validation, I don't think it's a big deal to have sanity-check warnings in assert-driven code. Just because a field can store a couple dozen digits doesn't mean that a flag shouldn't be raised when you see numbers more than 6-7 digits.
There are already similar checks in business code -- you can't sell a negative quantity at a cash register, you have to do a return. Operating systems make similar checks, asking for confirmation of "dangerous" or unusual situations (like permanently removing data.)
Why wouldn't a financial management/accounting system have similar rules enforced and monitored?
I read this article as Microsoft demanding that hundreds of thousands of customers install software upgrades and patches, rather than paying less than $9M for the right to keep delivering Office as-is.
So while they keep advertising about their IP "guarantee", when push comes to shove it seems they'd rather offload the expense to end-users and customers rather than deal with it properly.
And no, I don't care about the details or "validity" of the above patent. Point is that MS promises to protect the user, then rolls over when it looks like it might cost some money to do instead of talk.
You neglected another key point of the real question -- which company is going to be around next year? 2 years? 5? 10?
What if the interesting job is with a company that has no perceived revenue stream, a dot-bomb tech-driven business plan (whether they label it 2.0 or not), and no real business plan other than a hope to be bought out? Are you really that "interested" in finding another job when they start bouncing paycheques on their way to bankruptcy?
Unfortunately that's the real world of most startups. Great talk, great perks, low pay, long hours, no business future, no budget.
What, praytell, is the difference between a functional language like Haskell and a well-designed C++ template library? A few pure abstract classes acting as interface specifications, and you plug-and-play the implementations based on the data being transformed -- which is all that a functional language does behind the syntax and performance tweaks.
Make use of a factory interface for instantiating objects so the business programmers don't need to deal with the nitty-gritty of C++ syntax (which can be a tad typo-prone.)
Follow the ownership of the objects throughout your implementation. At any given time, one object owns another, with the owner responsible for releasing it's children before itself. Java J2EE still requires the concept of an explicit release/free to coordinate resources; only J2SE lets you more or less ignore deallocation.
Alternatively, impose a use counter or make use of one of the OSS instance management libraries out there.
As has been mentioned previously, test often, and make sure to do regression runs when making changes to the architectural approach.
Make heavy use of DEBUG-flagged assertions to state your assumptions about the data when code is executing. If those assertions go off in testing, either you're making bad assumptions or there is a problem with the data source. You should be able to disable the DEBUG code in production -- no DEBUG-flagged code should have side-effects.
Always validate remote API arguments. You mentioned a distributed application -- assume the peer is going to send you garbage, and deal with it.
Don't forget to unplug network cables randomly during load testing. It will happen in the real world. How does the cluster deal with that?
I thought about it for a while, and realized what construct a full join refers to. I've never needed one, but I think a three-way union would do it ( left+right, left+null, null+right ).
I'm a little confused. Wouldn't *=* just be an unqualified select matrix? (I forget the math/set theory term for the construct, but for n rows in one table and m in another, you get a result set of n * m elements.)
I believe the same is true of Sybase ASE 12.5 -- left and right outer ANSI join syntax is supported.
I also believe that 30 character limit might have something to do with the idea of "standard". Some databases extend the standard, but that doesn't mean a good database design assumes the database will do so. Besides, why in the world would you need more than 30 characters for a DDL object?
IIRC the standard used to be either 18 or 24 characters, so 30 is actually pretty generous in comparison.
For it to be entrapment, someone would have had to approach him with an offer to buy the stolen source code. He posted an offer to sell the source code on a website, so he initiated the exchange.
Slashdot Mirror
Here's a kicker.
Sony goes to all these DRM efforts to stop you from ripping CDs to use on your other media players.
Sony sells their ATRAC3/MP3 player with SonicStage CD ripping software so you cna use your existing CDs with your media player.
WTF? Which way do you want it Sony? Maybe you should figure out your business plan before your RIAA music division keeps on with their nonsense.
Ah, I see. I point out that it's a two-sided coin and call those who want it both ways wankers, and that's trolling. :P
It doesn't change the truth -- you can't have it both ways. Either security works and you have privacy/protection, or you can pirate to your heart's content. The writing is on the wall: society and legislation want the privacy.
The objectives have not changed, people are just beginning to understand that if I protect my information, no one else can access it. The reverse side of that equation is that if someone protects their information, I can't access it.
i.e. The technology that allows the *AA to come up with usage-restriction schemes is the same technology being mandated by privacy advocates when your address, financial, health care, and other information is accessed. It is expressly stated in the Canadian requirements for such systems that you must log every attempt to access a document.
So stop your freakin' panicking already, you anti-DRM wankers! Take it to the courts and fight the problem where it needs to be solved -- on a social/legal level.
That includes whoever came up with this flamebait phrasing:
<sarcasm> Yeah, but they don't wear brown shirts, red and black armbands, or thump the Koran this time. This time it's Americans thumping a puritan Bible, so it's ok, right? </sarcasm>
Haven't they already extended the limits on copyright to satisfy Disney's desire to retain control over a certain cartoon mouse?
No, the **AA does their best to force me to buy replacement media for as many playback device formats as I have access to.
The auto manufacturers, on the other hand, produce vehicles that last long enough to keep the customer satisfied if they put in the maintenance. They'd like you to buy more often, but they know that if their vehicles don't last, no one will buy them.
The **AA bully and intimidate everyone from barely-teens to aging grandmothers. They rely on lobbyists and bought politicians to "earn" their ever-increasing legal "rights" at the expense of what used to be consumer rights.
No business is guaranteed a market. If your market dries up and you don't find a way to adapt, you join the infamous buggy-whip maker. I'm sure they tried to push for legislation to make the "motor car" illegal, too.
The **AA has spent the past 20 years trying to change the rules. You used to be able to send back cassettes or albums for replacements when damaged; the only charge was shipping.
Now they tell you it can't be replaced, because that version has been replaced by a "new" release, even with relatively-recently purchased media.
Currently they're trying to cut it back further, so that it's not even legal for you to listen to your media on a portable device without paying yet again.
To hell with the greedy bastards. Once or twice at the trough was more than enough -- no more.
I actually buy a copy of Linux (usually SuSE) when I do an upgrade. Sure I could just download a couple .iso's and it's all legal, but I figure it's only fair to spend a few dollars when upgrading.
Buying a copy means some meagre portion of a developer's salary is paid to keep working on patches, updates, and integration of the overall system. For all the talk about open source, the truth is that commercial and OSS vendors pay for a lot of skilled developer resources to maintain the common source code.
Vendors provide support services, whether their code base is OSS or not.
A lot of industries also lend developer expertise, fixing bugs that affect the specific platform/tool combinations they're dealing with. Someone has to do it -- and a business needs it badly enough to pay to have it done.
Is it possible that this one-copy buyer sees OS/X as mostly OSS with a few GUI applications and yet another shiny desktop? How much is a desktop with utilities worth nowadays?
I think fine tuning would be the main benefit.
My understanding is that mainframe DB/2 lets you specify the type of storage and indexing to provide over data. PostgreSQL provides some control over that detail. MySQL was pretty much built as a virtual database over real containers. Sybase and Ingres have always allowed a fair bit of control over their indexing and container options. Oracle has some tuning options as well.
But if a vendor is targetting environments that need fine-tuning to eek every last bit of performance out of the hardware, it's always good to have more options for certain performance profiles. They're not common, but there might be a large enough customer base to consider adding that type of tuning to Oracle.
The fine-tuning options are all that significantly distinguish the top 3-4 RDBMS vendors. They all meet the checklists for multi-language, multi-platform drivers, SQL92+ standards support, integrity constraints, triggers, stored procs, replication options, online backup, and hot-failover and/or cluster support. You just need to buy into vendor-specific syntax to tweak the performance. It's getting hard to find specific areas where one vendor is "better" or "easier" than another.
The article may be a troll, but I notice no one is actually talking about network operating systems in response.
J2EE is a network OS programming kit. DotNet is another. QNX was designed as a networked/RPC version of *nix.
Given the market share the first two of that last have, I think the argument about whether network-bootable OS components/updates are viable is a moot point -- they've been here for years. Someone else just finally realized the fact.
The very first iteration of what eventually became Unix was a simple task switcher to allow a game to run at the same time as actual work. Technically it wasn't multi-user, because there was only a system console.
Actually I'm not talking about reject rules, I'm talking about confirmation-required flagging. So the only time an operator has to intervene is if flags are getting raised while processing a particular piece of data.
In some environments, there might even be multiple digital signatures required to approve such abnormal requests.
Some businesses just consider such checks and balances to be essential for fraud prevention and detection, but they can be used to catch a lot of other potentially expensive data problems early.
Not to defend outright piracy, but I know many people (sometimes myself as well) who rent and return movies within 24 hours. It only takes a couple hours to watch a movie, and it beats the heck out of TV tripe most of the time. Some of them will watch a couple movies a night.
So for them, this "heavy usage pattern" is perfectly normal and not a sign of piracy. Having their shipments "throttled" means they aren't getting the full use of the service they signed up for. It's up to the service provider to adjust capacity and/or pricing tiers to deal with the load, not choke off paid subscriptions that actually use the service as advertised.
Compared to the effort and expense of doing data range and argument validation, I don't think it's a big deal to have sanity-check warnings in assert-driven code. Just because a field can store a couple dozen digits doesn't mean that a flag shouldn't be raised when you see numbers more than 6-7 digits.
There are already similar checks in business code -- you can't sell a negative quantity at a cash register, you have to do a return. Operating systems make similar checks, asking for confirmation of "dangerous" or unusual situations (like permanently removing data.)
Why wouldn't a financial management/accounting system have similar rules enforced and monitored?
I read this article as Microsoft demanding that hundreds of thousands of customers install software upgrades and patches, rather than paying less than $9M for the right to keep delivering Office as-is.
So while they keep advertising about their IP "guarantee", when push comes to shove it seems they'd rather offload the expense to end-users and customers rather than deal with it properly.
And no, I don't care about the details or "validity" of the above patent. Point is that MS promises to protect the user, then rolls over when it looks like it might cost some money to do instead of talk.
The code was designed, exercised, tested, and executed properly from what you're saying. The display failed due to hardware problems.
In what way is that hardware failure related to code coverage or any other form of software testing or QA metric?
You neglected another key point of the real question -- which company is going to be around next year? 2 years? 5? 10?
What if the interesting job is with a company that has no perceived revenue stream, a dot-bomb tech-driven business plan (whether they label it 2.0 or not), and no real business plan other than a hope to be bought out? Are you really that "interested" in finding another job when they start bouncing paycheques on their way to bankruptcy?
Unfortunately that's the real world of most startups. Great talk, great perks, low pay, long hours, no business future, no budget.
No irony -- functional languages have baffled me for a long time because they seem to be syntactic sugar for template/interface programming.
In other words crafting stable C++ is as much about process, code samples, and enforcing standards as it is about selecting the right tools.
What, praytell, is the difference between a functional language like Haskell and a well-designed C++ template library? A few pure abstract classes acting as interface specifications, and you plug-and-play the implementations based on the data being transformed -- which is all that a functional language does behind the syntax and performance tweaks.
Make use of a factory interface for instantiating objects so the business programmers don't need to deal with the nitty-gritty of C++ syntax (which can be a tad typo-prone.)
Follow the ownership of the objects throughout your implementation. At any given time, one object owns another, with the owner responsible for releasing it's children before itself. Java J2EE still requires the concept of an explicit release/free to coordinate resources; only J2SE lets you more or less ignore deallocation.
Alternatively, impose a use counter or make use of one of the OSS instance management libraries out there.
As has been mentioned previously, test often, and make sure to do regression runs when making changes to the architectural approach.
Make heavy use of DEBUG-flagged assertions to state your assumptions about the data when code is executing. If those assertions go off in testing, either you're making bad assumptions or there is a problem with the data source. You should be able to disable the DEBUG code in production -- no DEBUG-flagged code should have side-effects.
Always validate remote API arguments. You mentioned a distributed application -- assume the peer is going to send you garbage, and deal with it.
Don't forget to unplug network cables randomly during load testing. It will happen in the real world. How does the cluster deal with that?
And the list goes on...
I wasn't actually talking about full joins for two of those limbs.
left+right is an obvious equality join.
left+null is left with null values for the right columns, where left is not in right.
null+right is right with null values for the left columns, where right is not in left.
That gives the database simpler constructs to evaluate while preparing the result set than a pair of outer joins.
OTOH a decent optimizer should take either form and end up doing the same internal operations.
I thought about it for a while, and realized what construct a full join refers to. I've never needed one, but I think a three-way union would do it ( left+right, left+null, null+right ).
I'm a little confused. Wouldn't *=* just be an unqualified select matrix? (I forget the math/set theory term for the construct, but for n rows in one table and m in another, you get a result set of n * m elements.)
I believe the same is true of Sybase ASE 12.5 -- left and right outer ANSI join syntax is supported.
I also believe that 30 character limit might have something to do with the idea of "standard". Some databases extend the standard, but that doesn't mean a good database design assumes the database will do so. Besides, why in the world would you need more than 30 characters for a DDL object?
IIRC the standard used to be either 18 or 24 characters, so 30 is actually pretty generous in comparison.
For it to be entrapment, someone would have had to approach him with an offer to buy the stolen source code. He posted an offer to sell the source code on a website, so he initiated the exchange.