Re:All this time I though...
on
The Zen of SOA
·
· Score: 1
Tell me, do any of you have an architecture that is not 'Service Oriented', and if so, how do you use it, if your architecture isn't designed to accommodate/enable 'services' (i.e. functionality), what is its purpose.
There's architecture which is "building oriented" and which supports services like "electric power" and "drains". Sometimes it gets a bit bloated, but mostly it's not too bad once deployed.
Re:Single provider and SOA?
on
The Zen of SOA
·
· Score: 1
Yes, although this is more about services interacting with each other, something that Unix is not too good on (named pipes only go so far). Unix (or at least the scripting part of it) is more like command line tools working together, mostly on the same box. SOA's are more indifferent on where the services are run.
Trust me in this, the Unix way is better than you give it credit for. In particular, pipes are rather more versatile and strings much more so. And people have been doing coupled local-remote scripts for many years. (Heck, I was taught how to do that as an undergraduate, many years ago...) Some of the tools we use have changed (e.g. ssh instead of rsh) but the principles have not. As for location-independence of services, I can remember using such things back in 1992. They sucked because the implementation was fluffed (a lack of modelling of the service usage pattern and the properties of the information system meant that it actually failed to do what it should in effectively the worst possible way!) but the system demonstrated that such things could be done.
What SOA does do is greatly increase the complexity of the overall system. When everything is working, that's not a big deal. When things go wrong, it gets "interesting". The cascading catastrophic failure is more common than it ought to be. The more service providers you have in the mix, the more the complexity goes up, and the harder it is to do well.
Oh, and most commercial providers still like to clap their hands together three times and declaim that they believe in fairies^Wlock-in.
I heard he mined the minerals and metals himself. Smelted and forged them. Used his own personal laser pointer to hand cut every chip. Speaks in code to a special computer that compiles it on the fly. And then hand assembles each and every Apple devise.
How will they carry on without him?
Well, they could try calling in Bruce Schneier to consult...
Actually, since it's a time of war and in a manner prescribed by law, requiring residents to quarter soldiers would be constitutional.
But throwing the residents in prison for not quartering soldiers when that's due to a shortage of soldiers, that would qualify for the "lunatic" award...
This also helps with cooling because the room AC/DC converter can be cooled with a dedicated system, either liquid, or part of the HVAC system.
Not really. Take any random computer (especially a server) and the heat generated by the power supply is usually quite a bit less than that generated by the CPU and hard disks. (Put your hand behind a power supply fan and it'll feel quite warm but remember that fan is sucking hot air out of the case that wasn't necessarily generated by the power supply.) The only time a computer power supply really starts to heat up is if it's being pushed to its limit or is highly inefficient, both of which are error conditions rather than normal operation.
Right now, the limit on big datacenters is basically how fast you can push power in and heat out (there is a very intimate relationship between them!) Some things have to be in the room, but others can be outside, and power is relatively easy this way; we know how to reduce the resistance of cables (put two side by side is a classic solution) and getting the power supplies out of the machine room lets you optimize the cooling better, even if the power supplies aren't the #1 problem. (In fact, with power supplies if you're not in the machine room, you can actually think in terms of letting them run hotter and using purely passive convective cooling. Requires a really different design, but might be better.)
Datacenter design is one huge mass of compromises. We shouldn't be surprised at that.
(OK, many datacenters are really limited by availability of electrical power. A bit like a large aluminum smelter or other heavy industry I suppose.)
That idea seems rather unworkable. Companies will just pass the recycling cost on to the consumer; there will be no savings. And the company has no particular way to enforce proper disposal; that needs to be done at a regulatory level. And finally what happens if the company should go bankrupt, or even be located in a different country?
At the regulatory level, the key thing is that the company should be made responsible. They may well subcontract the responsibility to a specialist recycler and pass the costs on to the customer; that's OK. The regulation is required to make sure that firms don't weasel out of doing it.
No, the cost belongs with the owner.
Once the cost is with the manufacturer, the cost will be passed on to the consumer.
How does this relate to a resistor which undergoes a discontinuous resistance change under critical conditions? Can you explain how it relates to the advertised device? Where is the charge being stored? Please continue to assume that I'm stupid, and explain the reasoning. My electromagnetic theory is thirty five years in the past now.
No electronic component has truly discontinuous behavior under critical conditions. Even physical switches have rather complex transients (which is why they need a debounce circuit), and transistors are interesting analog devices. It's just that they're non-linear devices and (in computers) they're mostly used biased so that the circuits have (to a good approximation) binary behavior; the prerequisite for that is non-linearity.
Now, if there was an effect that was previously a theoretical one, or at best a lab curiosity, and HP have managed to figure out how to turn it into something practical, then more power to their elbows. It'll be very interesting to see what the electronic engineers of the world do with it.
If you say yes to either one, centralized computing or software subscriptions, you're actually saying yes to BOTH.
I think that tinfoil hat has managed to slip over your eyes.
For software that's happy on your desktop or laptop, the best place for it is on that desktop/laptop. And right now, the mass market tends to be not keen on rental software (outside gaming, where it seems to be somewhat workable).
Once you get to the heavier-weight stuff (top-end simulators of various kinds particularly) then prices to buy go up fast. (Why? Because they're difficult to write and the total market isn't that big. Even if the software was open source, it would still be expensive when all costs are accounted for, since the chance is that you'd have to retain a programmer to help.) And you need a lot of processing to support this sort of thing - computational fluid dynamics is ever so thirsty for CPU, and real-life datasets are gigantic - though the exact details of what is the best way to do it will vary according to the exact nature of the code and how it is used (some prefer old-school big-iron, some like high-availability clusters, some are happy with scavenged cycles from lots of systems). All that costs (well, unless you're lucky enough to persuade lots of people to help out like Folding@home, but that's unusual) and that means that the whole service shebang is expensive. And this sort of thing is actually a good candidate for a pay-for on-the-web service, because that lets the service provider take care of all the stupid shitty details of running the app on all sorts of weird setups, rather than forcing every noob through a horrible learning curve; some people like finding out such things, but they're a minority.
Doing rental access to this sort of thing makes a bunch of sense, but only if the price is right and the software vendor is on board. OTOH, for a small shop who just needs occasional access to the software, this is great. They buy access to what they need, and for less than it would have cost them otherwise. (Otherwise, they actually wouldn't have used simulation at all; they're not in the HPC business, and full licenses to the software that's any good would be a huge fraction of their turnover.)
In short, Software as Service can work, and so can centralized computing. But not for everything, and there needs to be a market of providers. Right now, the market (such as it is) is terribly dysfunctional, with most service-level agreements looking like bad jokes. (What is it with cloud service providers? Customers can tell if you're trying to give them the shaft...)
It sounds like all the enterprise software I've ever encountered - the smaller, more specialised systems have crappy UIs that make the program not much fun to use. Only the big guns like MS or Adobe seem to have the time/money/desire to design a useful interface.
GUIs take a lot of time and effort to build, especially the "intuitive" ones; there's just a lot of bits and pieces that have to be done. For example, you have to remember to handle all the different ways of "intuitive interaction" that users come up with, including all the ones that aren't intuitive to you...
If you're only selling a few copies (for lots each) and you can substitute training, why blow masses of cash on the GUI? It won't make you earn more. (Training is better because you can charge the customer for it. It's only with the mass market that that gets awkward.)
Just zero the disk once and odds are that will be more than good enough for any of your personal data, unless you are the fucking president or something.
The objection to GOTOs really is a kind of religious commandment (handed down by the Prophet Dijkstra), and it's every bit as logical as most religious commandments.
Mod parent up; he's nailed exactly what's going on.
I should note that what Dijkstra was really complaining about originally was using GOTOs for things that are really basic structured programming constructs: building your own WHILE with GOTO is (almost) always a bad idea. But then he was railing about it back in the days of FORTRAN and PL/1, so what do you expect...
The addition of single bool avoids both the specialized cleanup() function and the goto:
[...]
The problem with that is that it tends (in real code) to either greatly increase the depth of nesting of the code (making it harder to understand) or, worse yet, spray a great collection of various flags indicating the various types of cleanup required and what conditions have and haven't been checked. Making conditions more complex isn't a good plan at all for maintenance, since it increases the difference between the pseudo-code for the function (i.e. the level that you think about, without all the grotty code for handling failure modes and obscure cases) and the real code...
In TDD, you actually let the tests drive the development. You first write a test, then the simplest code that will satisfy it. Then add more tests/assertions and modify the code.
The problem with test suites is determining when you have tests that cover everything they need to. I suspect that it is actually a problem equivalent to constructing a formal proof of correctness of the specification and of conformance of the implementation to the spec. (I've always found it necessary to have some tests that are driven by knowledge of the implementation, just to ensure that tricky emergent issues stay fixed.)
I have used perl extensively adn one thing its OO is not is "complex". Bless is very easy to use, its a very simple model? It couldnt be easier.
Theoretically, yes. But it's really a bit deep for most folks. In practice, it would have been better to have given something less flexible but with more syntactic sugar, since then it would have been easier for more people to have used it.
Language design is a fine line between power and usability. (Well, it is for languages that have real merit; some are just weak and difficult, but they're really just curiosities.)
However, as bad as we may think it is here in the United States (compared to the pre-9/11 world), things are much worse in the United Kingdom. The rights of the Individual in the UK are enshrined in Common Law (i.e., customary law passed down through the ages), and not explicitly delineated in any sort of constitutional document.
You're somewhat confused. There is a collection of UK constitutional law, though it is in multiple documents. Part of that constitutional law is the Human Rights Act, which is actually distinctly relevant to the rights of individuals and how officialdom may deal with them. (The details do vary from the US, but to claim that there is no constitution or protections is factually incorrect.)
i saw somebody mention disney earlier on. am i correct in thinking, since walt disney died in 1966, according to current EU law, the images of characters such as micky mouse and donald duck will enter the public domain in december of 2036 in the EU (70 years after death) and 2026 in the USA (95 years after initial copyright)?
You have to differentiate between copyright and trademark. Copyright relates to the ability to make copies of "Steamboat Willie", display them in a public place, that sort of thing. Trademark relates to the ability to make wholly new Mickey Mouse cartoons, merchandise, etc. If the copyright on early Mickey Mouse cartoons has expired (not that I'm claiming that it has in your jurisdiction) that means you can create derivatives that are clearly from those specific works, but you'll have to be careful when you do so because you'll need to avoid trademark problems: trademarks are perpetual so long as they're defended actively (and Disney definitely aren't fools that way...)
I'd probably argue that the Pioneer craft edge out the rovers, but the rovers are an easy third.:)
There's also the two Voyagers; I remember being spellbound by what they found out about the outer planets.
In fact, when I think about it NASA have done some amazing scientific missions. We've learned so much about so many worlds, and in particular we're starting to see just how special Earth is. A real golden age for space science, and I don't think we're out of it yet...
In the computer room at my college, many years ago, there was the following sign:
Rule 1: Always make a backup. Rule 2: Always make a backup. (This is a backup of Rule 1)
Just because things are now on Web 2.0 services over the internet doesn't change the fundamental dictum. If you care about the data, it is you who needs backups. If you don't make backups, obviously you don't care (enough)...
Why not deploy your app on JRuby, and if there are still performance bottlenecks, write those parts in real Java ?
Apart from whether to use Java or not, this post exposes a really good point: if you've got performance problems, you're often best off rewriting part of your app in a lower-level language. Keep the rewritten part small though, since it's usually easier to work in a higher-level language. And always measure whether you're addressing a real bottleneck and making it better...
(The identities of the higher/lower-level language aren't so important as the principle of splitting.)
(it's probably safer if the government isn't managing this anyway)
Safer for the government bureaucrats who want to hide exactly what they're doing with the data? Definitely.
Safer for the people who need to keep government abuses in check? Absolutely not.
Whenever your leaders start outsourcing to the private sector or relocating operations to oppressive regimes be warned. They are outsourcing blame and accountability as well.
What you have to understand is the key advantage to outsourcing is that the government will then claim that they can't permit any kind of public oversight of the activity at all due to it being "commercially sensitive" or whatever the exact phrase is, despite the fact that it is the public that is ultimately the purchaser of the activity. The slippery swines have tried it before with the financing of public projects, and look where that is taking us. (OTOH, it's a good thing that it is hard to finance many public projects; it's stopped various local governments from doing a wide range of utterly idiotic things that the electorate wanted nothing to do with...)
Slashdot Mirror
What's wrong with the turtles?
Turtles can't do algebra.
But they can do geometry:http://en.wikipedia.org/wiki/Turtle_graphics
OK, I'm confused. Can turtles do Algebraic Geometry?
Tell me, do any of you have an architecture that is not 'Service Oriented', and if so, how do you use it, if your architecture isn't designed to accommodate/enable 'services' (i.e. functionality), what is its purpose.
There's architecture which is "building oriented" and which supports services like "electric power" and "drains". Sometimes it gets a bit bloated, but mostly it's not too bad once deployed.
Yes, although this is more about services interacting with each other, something that Unix is not too good on (named pipes only go so far). Unix (or at least the scripting part of it) is more like command line tools working together, mostly on the same box. SOA's are more indifferent on where the services are run.
Trust me in this, the Unix way is better than you give it credit for. In particular, pipes are rather more versatile and strings much more so. And people have been doing coupled local-remote scripts for many years. (Heck, I was taught how to do that as an undergraduate, many years ago...) Some of the tools we use have changed (e.g. ssh instead of rsh) but the principles have not. As for location-independence of services, I can remember using such things back in 1992. They sucked because the implementation was fluffed (a lack of modelling of the service usage pattern and the properties of the information system meant that it actually failed to do what it should in effectively the worst possible way!) but the system demonstrated that such things could be done.
What SOA does do is greatly increase the complexity of the overall system. When everything is working, that's not a big deal. When things go wrong, it gets "interesting". The cascading catastrophic failure is more common than it ought to be. The more service providers you have in the mix, the more the complexity goes up, and the harder it is to do well.
Oh, and most commercial providers still like to clap their hands together three times and declaim that they believe in fairies^Wlock-in.
That requires time.
No, switching from suck to blow requires the Schwartz! Everyone knows that...
I heard he mined the minerals and metals himself. Smelted and forged them. Used his own personal laser pointer to hand cut every chip. Speaks in code to a special computer that compiles it on the fly. And then hand assembles each and every Apple devise.
How will they carry on without him?
Well, they could try calling in Bruce Schneier to consult...
Actually, since it's a time of war and in a manner prescribed by law, requiring residents to quarter soldiers would be constitutional.
But throwing the residents in prison for not quartering soldiers when that's due to a shortage of soldiers, that would qualify for the "lunatic" award...
This also helps with cooling because the room AC/DC converter can be cooled with a dedicated system, either liquid, or part of the HVAC system.
Not really. Take any random computer (especially a server) and the heat generated by the power supply is usually quite a bit less than that generated by the CPU and hard disks. (Put your hand behind a power supply fan and it'll feel quite warm but remember that fan is sucking hot air out of the case that wasn't necessarily generated by the power supply.) The only time a computer power supply really starts to heat up is if it's being pushed to its limit or is highly inefficient, both of which are error conditions rather than normal operation.
Right now, the limit on big datacenters is basically how fast you can push power in and heat out (there is a very intimate relationship between them!) Some things have to be in the room, but others can be outside, and power is relatively easy this way; we know how to reduce the resistance of cables (put two side by side is a classic solution) and getting the power supplies out of the machine room lets you optimize the cooling better, even if the power supplies aren't the #1 problem. (In fact, with power supplies if you're not in the machine room, you can actually think in terms of letting them run hotter and using purely passive convective cooling. Requires a really different design, but might be better.)
Datacenter design is one huge mass of compromises. We shouldn't be surprised at that.
(OK, many datacenters are really limited by availability of electrical power. A bit like a large aluminum smelter or other heavy industry I suppose.)
Hey, people have survived worse things, like gcc version changes, Qt3 => Qt4, Gtk 1 => Gtk2...
Not to mention Perl4 -> Perl5.
Or Perl5 -> Perl6...
(The bane of my life has been glibc versioning. Brought to you by the "Stable ABI? What's that?" school of programming.)
This leads to chronic "cargo cult programming" (i.e. code, test, repeat) in many organizations.
If they're actually using testing, they're doing better than many shops...
That idea seems rather unworkable. Companies will just pass the recycling cost on to the consumer; there will be no savings. And the company has no particular way to enforce proper disposal; that needs to be done at a regulatory level. And finally what happens if the company should go bankrupt, or even be located in a different country?
At the regulatory level, the key thing is that the company should be made responsible. They may well subcontract the responsibility to a specialist recycler and pass the costs on to the customer; that's OK. The regulation is required to make sure that firms don't weasel out of doing it.
No, the cost belongs with the owner.
Once the cost is with the manufacturer, the cost will be passed on to the consumer.
Be careful what you ask for. No matter how bad your ISP is, it can't suck0rs as much as *vacuum* ISP.
Don't forget black-hole ISPs. They suck nearly as much as... hmm, what a choice...
How does this relate to a resistor which undergoes a discontinuous resistance change under critical conditions? Can you explain how it relates to the advertised device? Where is the charge being stored? Please continue to assume that I'm stupid, and explain the reasoning. My electromagnetic theory is thirty five years in the past now.
No electronic component has truly discontinuous behavior under critical conditions. Even physical switches have rather complex transients (which is why they need a debounce circuit), and transistors are interesting analog devices. It's just that they're non-linear devices and (in computers) they're mostly used biased so that the circuits have (to a good approximation) binary behavior; the prerequisite for that is non-linearity.
Now, if there was an effect that was previously a theoretical one, or at best a lab curiosity, and HP have managed to figure out how to turn it into something practical, then more power to their elbows. It'll be very interesting to see what the electronic engineers of the world do with it.
If you say yes to either one, centralized computing or software subscriptions, you're actually saying yes to BOTH.
I think that tinfoil hat has managed to slip over your eyes.
For software that's happy on your desktop or laptop, the best place for it is on that desktop/laptop. And right now, the mass market tends to be not keen on rental software (outside gaming, where it seems to be somewhat workable).
Once you get to the heavier-weight stuff (top-end simulators of various kinds particularly) then prices to buy go up fast. (Why? Because they're difficult to write and the total market isn't that big. Even if the software was open source, it would still be expensive when all costs are accounted for, since the chance is that you'd have to retain a programmer to help.) And you need a lot of processing to support this sort of thing - computational fluid dynamics is ever so thirsty for CPU, and real-life datasets are gigantic - though the exact details of what is the best way to do it will vary according to the exact nature of the code and how it is used (some prefer old-school big-iron, some like high-availability clusters, some are happy with scavenged cycles from lots of systems). All that costs (well, unless you're lucky enough to persuade lots of people to help out like Folding@home, but that's unusual) and that means that the whole service shebang is expensive. And this sort of thing is actually a good candidate for a pay-for on-the-web service, because that lets the service provider take care of all the stupid shitty details of running the app on all sorts of weird setups, rather than forcing every noob through a horrible learning curve; some people like finding out such things, but they're a minority.
Doing rental access to this sort of thing makes a bunch of sense, but only if the price is right and the software vendor is on board. OTOH, for a small shop who just needs occasional access to the software, this is great. They buy access to what they need, and for less than it would have cost them otherwise. (Otherwise, they actually wouldn't have used simulation at all; they're not in the HPC business, and full licenses to the software that's any good would be a huge fraction of their turnover.)
In short, Software as Service can work, and so can centralized computing. But not for everything, and there needs to be a market of providers. Right now, the market (such as it is) is terribly dysfunctional, with most service-level agreements looking like bad jokes. (What is it with cloud service providers? Customers can tell if you're trying to give them the shaft...)
It sounds like all the enterprise software I've ever encountered - the smaller, more specialised systems have crappy UIs that make the program not much fun to use. Only the big guns like MS or Adobe seem to have the time/money/desire to design a useful interface.
GUIs take a lot of time and effort to build, especially the "intuitive" ones; there's just a lot of bits and pieces that have to be done. For example, you have to remember to handle all the different ways of "intuitive interaction" that users come up with, including all the ones that aren't intuitive to you...
If you're only selling a few copies (for lots each) and you can substitute training, why blow masses of cash on the GUI? It won't make you earn more. (Training is better because you can charge the customer for it. It's only with the mass market that that gets awkward.)
Just zero the disk once and odds are that will be more than good enough for any of your personal data, unless you are the fucking president or something.
"Can you guys recover my data?"
"Yes we can!"
The objection to GOTOs really is a kind of religious commandment (handed down by the Prophet Dijkstra), and it's every bit as logical as most religious commandments.
Mod parent up; he's nailed exactly what's going on.
I should note that what Dijkstra was really complaining about originally was using GOTOs for things that are really basic structured programming constructs: building your own WHILE with GOTO is (almost) always a bad idea. But then he was railing about it back in the days of FORTRAN and PL/1, so what do you expect...
The addition of single bool avoids both the specialized cleanup() function and the goto:
[...]
The problem with that is that it tends (in real code) to either greatly increase the depth of nesting of the code (making it harder to understand) or, worse yet, spray a great collection of various flags indicating the various types of cleanup required and what conditions have and haven't been checked. Making conditions more complex isn't a good plan at all for maintenance, since it increases the difference between the pseudo-code for the function (i.e. the level that you think about, without all the grotty code for handling failure modes and obscure cases) and the real code...
In TDD, you actually let the tests drive the development. You first write a test, then the simplest code that will satisfy it. Then add more tests/assertions and modify the code.
The problem with test suites is determining when you have tests that cover everything they need to. I suspect that it is actually a problem equivalent to constructing a formal proof of correctness of the specification and of conformance of the implementation to the spec. (I've always found it necessary to have some tests that are driven by knowledge of the implementation, just to ensure that tricky emergent issues stay fixed.)
I have used perl extensively adn one thing its OO is not is "complex". Bless is very easy to use, its a very simple model? It couldnt be easier.
Theoretically, yes. But it's really a bit deep for most folks. In practice, it would have been better to have given something less flexible but with more syntactic sugar, since then it would have been easier for more people to have used it.
Language design is a fine line between power and usability. (Well, it is for languages that have real merit; some are just weak and difficult, but they're really just curiosities.)
However, as bad as we may think it is here in the United States (compared to the pre-9/11 world), things are much worse in the United Kingdom. The rights of the Individual in the UK are enshrined in Common Law (i.e., customary law passed down through the ages), and not explicitly delineated in any sort of constitutional document.
You're somewhat confused. There is a collection of UK constitutional law, though it is in multiple documents. Part of that constitutional law is the Human Rights Act, which is actually distinctly relevant to the rights of individuals and how officialdom may deal with them. (The details do vary from the US, but to claim that there is no constitution or protections is factually incorrect.)
i saw somebody mention disney earlier on. am i correct in thinking, since walt disney died in 1966, according to current EU law, the images of characters such as micky mouse and donald duck will enter the public domain in december of 2036 in the EU (70 years after death) and 2026 in the USA (95 years after initial copyright)?
You have to differentiate between copyright and trademark. Copyright relates to the ability to make copies of "Steamboat Willie", display them in a public place, that sort of thing. Trademark relates to the ability to make wholly new Mickey Mouse cartoons, merchandise, etc. If the copyright on early Mickey Mouse cartoons has expired (not that I'm claiming that it has in your jurisdiction) that means you can create derivatives that are clearly from those specific works, but you'll have to be careful when you do so because you'll need to avoid trademark problems: trademarks are perpetual so long as they're defended actively (and Disney definitely aren't fools that way...)
I'd probably argue that the Pioneer craft edge out the rovers, but the rovers are an easy third. :)
There's also the two Voyagers; I remember being spellbound by what they found out about the outer planets.
In fact, when I think about it NASA have done some amazing scientific missions. We've learned so much about so many worlds, and in particular we're starting to see just how special Earth is. A real golden age for space science, and I don't think we're out of it yet...
In the computer room at my college, many years ago, there was the following sign:
Rule 1: Always make a backup.
Rule 2: Always make a backup. (This is a backup of Rule 1)
Just because things are now on Web 2.0 services over the internet doesn't change the fundamental dictum. If you care about the data, it is you who needs backups. If you don't make backups, obviously you don't care (enough)...
Why not deploy your app on JRuby, and if there are still performance bottlenecks, write those parts in real Java ?
Apart from whether to use Java or not, this post exposes a really good point: if you've got performance problems, you're often best off rewriting part of your app in a lower-level language. Keep the rewritten part small though, since it's usually easier to work in a higher-level language. And always measure whether you're addressing a real bottleneck and making it better...
(The identities of the higher/lower-level language aren't so important as the principle of splitting.)
(it's probably safer if the government isn't managing this anyway)
Safer for the government bureaucrats who want to hide exactly what they're doing with the data? Definitely.
Safer for the people who need to keep government abuses in check? Absolutely not.
Whenever your leaders start outsourcing to the private sector or relocating operations to oppressive regimes be warned. They are outsourcing blame and accountability as well.
What you have to understand is the key advantage to outsourcing is that the government will then claim that they can't permit any kind of public oversight of the activity at all due to it being "commercially sensitive" or whatever the exact phrase is, despite the fact that it is the public that is ultimately the purchaser of the activity. The slippery swines have tried it before with the financing of public projects, and look where that is taking us. (OTOH, it's a good thing that it is hard to finance many public projects; it's stopped various local governments from doing a wide range of utterly idiotic things that the electorate wanted nothing to do with...)