Beakman's World did a pretty good demonstration about this, and you may be able to do an adaptation of that approach:
You need things:
A collection of Nerf balls (or other soft small throwable objects).
A hoop of some sort.
A whiteboard and dry-erase markers.
A person to play the CPU.
A person to play the Engineer.
On the whiteboard you write the Requirements:
Design a program to get the CPU to throw a ball through the hoop.
The CPU is only allowed to execute instructions that are written on a whiteboard by the Engineer.
The CPU is to be as literal as possible when interpreting the instructions - so if the instructions are:
Throw ball at hoop
Then the CPU should pantomime throwing something, because the Program does not include the instruction "Pick up the ball".
Start with just the "Requirements" on the whiteboard. As the kids come up, you explain the rules, and let them tell you what instructions to write down.
You can even have a bit more fun - if a kid starts suggesting your standard juvenile crap ("Have him scratch his butt <snicker>") you can pull a Donald - "You have been caught goofing off at work - YOU'RE FIRED!".
If you are real meanie, you can even do more of the experience:
The PHB, who randomly changes the spec
The Hardware Engineer, who designs the hardware ("OOPS! - the CPU cannot pick up the balls because the arm is too short! Hardware Revision!")
the Vendor ("Oh No! The balls have gone End-Of-Life, and we have to find a replacement!")
The Marketing Guy ("I can't sell this doody - make it sing too!")
The Auditor ("You have failed to use the proper font for your program - your ISO-9000 certification is revoked - unless you want to hire me as a consultant to help you be in compliance....")
And the funny thing is, that it is incidents like this that cause people like me to not want to register with sites like the Times, precisely to prevent accidental disclosure of any information.
And yes, I know that you can lie on the forms, but my attitude is, why encourage sites to use registration by lying?
OK, before going into any of the "How" questions (How do I gather this information? How do I assure its validity?) let us as the more fundamental question:
Why do you want to gather this information?
For example, why do you care what browser and OS I am running? Unless you are selling computer software or hardware, you shouldn't give a rat's testicles what my computer is - so why do you need to know?
Why do you care where I am located? Are you trying to guess what the shipping costs will be? Are you trying to gauge whether your site's content is applicable or legal in my area?
Are you really sure this information is going to do you any good, even if you can aquire it?
And having aquired it, how are you going to insure that anything you've found out will NOT be "leaked" to somebody else? Considering the ChoicePoint et. al. fiascos, *I* would not want to gather one bit of information more than I absolutely needed - information you do not have you cannot leak.
But you see, Zonk had duped Rob. Had it been Rob who had duped Zonk, then the story would not have been removed. But since Zonk is lower on the totem pole than Rob, Zonk's story (and all comments) got Zonked.
A pity - one of my comments was lost there as well.
I wonder if they will now search the database and remove the comments from each poster's comment log to furthur hide the evidence.
Creating a dedicated Wine user and running all Wine apps as that user
Properly securing your base system (e.g. making sure all system directories are writable only by root, and not running as root.
chroot'ing the Wine install.
Configuring SELinux security tags to restrict Wine apps.
None of which have anything to do with the blitherings of this troll.
The sad thing is that some of this moron's follow-ups are still at +3, and that no matter what, the moron moderators who gave him positive points will, in all likelyhood, be metamoderated as "fair".
Xen does run windows. The developers used shared source and are not allowed to release the code.
If they cannot release the code so that others can use it, then the fact that they have Windows working under Xen is not terribly relevant to the discussion at hand.
You keep parroting the line "separation of state" - I do not think you know what that term means.
Prove me wrong - give me a clear description of what you mean by it, and how it applies to running Wine inside a virtualized environment vs. running Wine as a process in a a non-virtualized environment.
You don't install Wine into a virtual machine any more than you install Office or HalfLife into a virtual machine.
You install an OPERATING SYSTEM into a virtual machine, then you install applications on that OS.
Wine is an application, no different than OpenOffice. It uses the services of the underlying operating system to do its job. The fact that its job is to provide the APIs of a foreign operating system is incidental.
So, all that running Xen would do is to allow you to have an install of Linux or *BSD solely to run Wine - which would provide no real benefit to running Wine.
The only way in which Xen would be of use in running Windows programs would be if Windows ran under Xen - which last time I checked it DOES NOT.
The poster of the parent post is just trolling for stupid moderators, and obviously has already found at least one.
That Sony is going toward a (albeit stripped down) version of OpenGL as their graphics library is interesting.
We already know that Linux runs on several Sony platforms. We know that Sony isn't really all that comfortable with the idea of doing business with Microsoft, especially now that MS is a competitor with them in the gaming front.
Sony also has pretty good manufacturing and distribution capacity.
So, consider the following scenario:
Sony releases a PCI-Express version of their latest graphics chipset for their games, along with Windows drivers, and releases full programming specs to X.org to enable writing a Free Driver.
Sony releases video boards based upon said chip.
Sony says to developers, "OK, you can develop for Direct-X, and run on Windows and X-Box. Or you can develop for OpenGL and SDL, and run on Windows, Mac, Linux, and Playstation. Your choice."
Google has made exceptions to its cloaking rule in the past. For example, its crawler views password-protected content in the Google Scholar program that cannot be viewed by broad Web users, and indexes National Public Radio audio transcripts that are unavailable to Web users, Sullivan said.
Now, take this information along with the earlier issue of the new customization on the news.google.com site, which frequently lists news sites that require registration.
Those sites serve out different content for the Googlebot than they do for my browser, but obviously Google "makes an exception" in their case.
And that would be fine by me, if I have the option to disable reporting of such sites in my news.google.com cookie.
Please take this in the spirit of constructive criticism - if/. had a means for one user to send another a private message via the/. message system I would have used that method.
"I could care less about Sports...."
The expression is "I could not care less" - if you can care less, then you care some - if you cannot care less, then you don't care at all.
Now, if you want, you can use this expression:
I could care less about <thing>, but it would be an effort.
A select can only tell you that a given write descriptor can accept one byte or more. It does not guarantee that a write to that FD won't block.
So, you have to make all your outbound descriptors O_NOBLOCK and add code to check each write to verify the amount of data written on each write, which adds overhead.
Additionally, in an SMP machine, using one big loop plus a select prevents the operations from being spread across the processors.
Lastly, by multithreading the task, you greatly simplify the code as you do NOT have to be constantly checking for short reads/writes - and all that checking has overhead.
The cost of a thread switch is very small - this thinking that "threads are expensive, context switches are expensive" dates from a long-gone age.
Lastly - you DID notice that little bit about "I've experimentally verified this", did you? I can demonstrate that the previous version of Bittorrent would block downloads if the uploads were being throttled, and that by cutting back the upload bandwidth that BT tried to use to a pathetic trickle of the TC rate limit the download rates were restored to full speed, demonstrating that it was the blocking of the uploads that was throttling the downloads.
I wonder if the new version has multithreaded the client so that receive and transmit are in separate threads?
This is important if you are using traffic shaping on your upstream connection, as I am. I'm on ADSL, and so my upstream bandwidth is less than my downstream. To prevent BT from consuming all my upstream bandwidth I am using the tc module in the kernel to restrict the BT packets (the rate limiting in BT is next to useless, as each instance of the client will use the programmed bandwidth - there is no "global" sharing of the bandwidth, so if you have 4 clients running it will take 4 times the bandwidth of 1 client).
The problem is that if the client is blocked sending an outbound torrent packet (because the traffic shaper queue is full), the client will not process any available incoming data packets, and this will hammer the download speed - I have expermimentally verified this.
Now, if there were separate threads for downloading and for uploading, the uploading threads would block as the TC queue filled, but the download threads would not be blocked, and could handle the download at full speed.
If there are separate channels for uplink and downlink, this is not simulating full duplex, it IS full duplex.
However, this makes me wonder about something. Compare this with another wireless mesh network - AX.25 (amatuer radio packet). In AX.25, rather than an end-to-end ack, each node acks the packet upon receipt, then forwards the packet to the next hop - thus acks need only move one hop.
Now, while this does increase the complexity of each node's processing (for example, how do you handle it when a packet has been accepted by a node, but that node cannot deliver the packet to the next hope), it cuts the delay on moving freight.
Now, TCP has a means to work around this (the TCP ack window), but it makes me wonder - would it be beneficial to a node system like this to have a protocol-layer store and forward system?
For example, what if each node ran a HTTP cache, and when a client requests a page, each node in the chain from client to server buffered the data so that any drop-outs and/or turnaround delays would have a minimal effect on the transfer?
I wonder which would have the greater effect upon throughput - full duplex, or caching? And would both have an even greater effect?
Or, now that I think about it, a flying game wherein the tilt of the laptop controls the yoke of the aircraft - imagine the fun of sitting in a cybercafe holding your Powerbook in your hands and going "Vrroommm! PowPowPow!"
Or rather, the fun of watching somebody else doing that....
You should quit when you have your next job lined up.
That is, when you have the offer of employment from your new employer, and a starting date set.
I had a friend who did the "take this job and shove it" trick with what was truely a bad situation. However, it was several months before he had another job lined up, and he very nearly had to file for bankrupcy. It *did* screw his credit up for a long time, due to the amount of debt he racked up during that time.
All jobs suck - but some more than others.
So you should ask yourself, "Realistically, does this job suck worse than any new job I might get?"
Assuming the answer is "HELL YES!", then start looking for a new job - BUT DON'T LET YOUR CURRENT EMPLOYER KNOW. Make sure you tell any headhunters you work with that you don't want your current employer contacted.
Look long and well - do everything you can do to insure that your new job will suck less than your current job.
Then, when they offer you a position, set your start date no earlier than two and a half weeks into the future, get the formal (and legally binding) letter of offer and your letter of acceptance.
THEN, and ONLY then, do you go to your current boss and tender your resignation. And no matter how strong the temptation, no matter what you feel your justification is, no matter how badly you'd like to tell them off, resign in a calm, professional manner. This world is too damn small to say "First of all, you ain't no good, never been no good, you smell like old wet cheese, you pay shit...." - those words WILL come back to haunt you (like, the *next* time you go to look for a new job, and prospective employers are calling this guy!) Make sure you give them your two weeks (they may offer to let you go immediately or ask that you continue to work - be ready either way).
Also, when changing jobs, you are shaking your world up - so do your best to save up some emergency money before hand, and even if your new job pays 4x what you were making - act as though you were making your old salary and save the difference - at least for a year. Remember, last in, first out.
You may want to quit today - you may go home every night grinding your teeth, but USE that anger to drive your job search - remember, while your current job may suck, imagine how much MORE it will suck if you have to go crawling back in order to keep a roof over your head!
Craiglist: the 1990's called, and they want their dot-bomb business practices back.
Seriously - when I hear of a company doing dain-bramaged things like this, I think to myself "Well, there's a company with more money than common sense (or business sense)."
And of course, companies with more money than sense spontaniously decay, via moron emission, into companies with no cash and high debt, which then decay, via bankrupcy, into crap at auction.
Don't take the absence of any moral obligation as prohibition.
In other words, just because a game designer/movie director/author/actor/... is not obligated to place any moral values into their story does not mean they should be prohibited from doing so if they so choose.
For example - I like the Myth series by Robert Lynn Asprin, precisely because RLA does put good "life lessions" into the stories.
Too many slashbots will jump on this story saying "Yeah, those bastards ought not be allowed to put their morality into their work!"
When the real point is "They ought not be REQUIRED to put SOMEBODY ELSE'S morality into their work."
I am not speaking of hunting Vulpes vulpes, but rather the sport practiced by amatuer radio operators of tracking a hidden transmitter with radio direction finding equipment.
The proper place for information like language preference is not on the card, but rather in the bank's database that the ATM accesses.
Ideally, when the card is first inserted the ATM will ask for non-secure data from the bank - things like language pref and such. If the card is NOT valid, the bank could send back default data (to prevent using that to ease checking of forged cards).
By seperating the prefs from the card, you can update the card without losing the prefs.
(Slashbots: Notice that the word is losing, not loosing!)
Slashdot Mirror
You need things:
On the whiteboard you write the Requirements:
The CPU is only allowed to execute instructions that are written on a whiteboard by the Engineer.
The CPU is to be as literal as possible when interpreting the instructions - so if the instructions are:
Then the CPU should pantomime throwing something, because the Program does not include the instruction "Pick up the ball".
Start with just the "Requirements" on the whiteboard. As the kids come up, you explain the rules, and let them tell you what instructions to write down.
You can even have a bit more fun - if a kid starts suggesting your standard juvenile crap ("Have him scratch his butt <snicker>") you can pull a Donald - "You have been caught goofing off at work - YOU'RE FIRED!".
If you are real meanie, you can even do more of the experience:
And the funny thing is, that it is incidents like this that cause people like me to not want to register with sites like the Times, precisely to prevent accidental disclosure of any information.
And yes, I know that you can lie on the forms, but my attitude is, why encourage sites to use registration by lying?
OK, before going into any of the "How" questions (How do I gather this information? How do I assure its validity?) let us as the more fundamental question:
Why do you want to gather this information?
For example, why do you care what browser and OS I am running? Unless you are selling computer software or hardware, you shouldn't give a rat's testicles what my computer is - so why do you need to know?
Why do you care where I am located? Are you trying to guess what the shipping costs will be? Are you trying to gauge whether your site's content is applicable or legal in my area?
Are you really sure this information is going to do you any good, even if you can aquire it?
And having aquired it, how are you going to insure that anything you've found out will NOT be "leaked" to somebody else? Considering the ChoicePoint et. al. fiascos, *I* would not want to gather one bit of information more than I absolutely needed - information you do not have you cannot leak.
But you see, Zonk had duped Rob. Had it been Rob who had duped Zonk, then the story would not have been removed. But since Zonk is lower on the totem pole than Rob, Zonk's story (and all comments) got Zonked.
A pity - one of my comments was lost there as well.
I wonder if they will now search the database and remove the comments from each poster's comment log to furthur hide the evidence.
None of which have anything to do with the blitherings of this troll.
The sad thing is that some of this moron's follow-ups are still at +3, and that no matter what, the moron moderators who gave him positive points will, in all likelyhood, be metamoderated as "fair".
If they cannot release the code so that others can use it, then the fact that they have Windows working under Xen is not terribly relevant to the discussion at hand.
G.O.A.T.
You keep parroting the line "separation of state" - I do not think you know what that term means.
Prove me wrong - give me a clear description of what you mean by it, and how it applies to running Wine inside a virtualized environment vs. running Wine as a process in a a non-virtualized environment.
Parent post is pure, unadulterated bullshit.
You don't install Wine into a virtual machine any more than you install Office or HalfLife into a virtual machine.
You install an OPERATING SYSTEM into a virtual machine, then you install applications on that OS.
Wine is an application, no different than OpenOffice. It uses the services of the underlying operating system to do its job. The fact that its job is to provide the APIs of a foreign operating system is incidental.
So, all that running Xen would do is to allow you to have an install of Linux or *BSD solely to run Wine - which would provide no real benefit to running Wine.
The only way in which Xen would be of use in running Windows programs would be if Windows ran under Xen - which last time I checked it DOES NOT.
The poster of the parent post is just trolling for stupid moderators, and obviously has already found at least one.
That Sony is going toward a (albeit stripped down) version of OpenGL as their graphics library is interesting.
We already know that Linux runs on several Sony platforms. We know that Sony isn't really all that comfortable with the idea of doing business with Microsoft, especially now that MS is a competitor with them in the gaming front.
Sony also has pretty good manufacturing and distribution capacity.
So, consider the following scenario:
Sony releases a PCI-Express version of their latest graphics chipset for their games, along with Windows drivers, and releases full programming specs to X.org to enable writing a Free Driver.
Sony releases video boards based upon said chip.
Sony says to developers, "OK, you can develop for Direct-X, and run on Windows and X-Box. Or you can develop for OpenGL and SDL, and run on Windows, Mac, Linux, and Playstation. Your choice."
Discuss.
Now, take this information along with the earlier issue of the new customization on the news.google.com site, which frequently lists news sites that require registration.
Those sites serve out different content for the Googlebot than they do for my browser, but obviously Google "makes an exception" in their case.
And that would be fine by me, if I have the option to disable reporting of such sites in my news.google.com cookie.
"I could care less about Sports...."
The expression is "I could not care less" - if you can care less, then you care some - if you cannot care less, then you don't care at all.
Now, if you want, you can use this expression:
A select can only tell you that a given write descriptor can accept one byte or more. It does not guarantee that a write to that FD won't block.
So, you have to make all your outbound descriptors O_NOBLOCK and add code to check each write to verify the amount of data written on each write, which adds overhead.
Additionally, in an SMP machine, using one big loop plus a select prevents the operations from being spread across the processors.
Lastly, by multithreading the task, you greatly simplify the code as you do NOT have to be constantly checking for short reads/writes - and all that checking has overhead.
The cost of a thread switch is very small - this thinking that "threads are expensive, context switches are expensive" dates from a long-gone age.
Lastly - you DID notice that little bit about "I've experimentally verified this", did you? I can demonstrate that the previous version of Bittorrent would block downloads if the uploads were being throttled, and that by cutting back the upload bandwidth that BT tried to use to a pathetic trickle of the TC rate limit the download rates were restored to full speed, demonstrating that it was the blocking of the uploads that was throttling the downloads.
I wonder if the new version has multithreaded the client so that receive and transmit are in separate threads?
This is important if you are using traffic shaping on your upstream connection, as I am. I'm on ADSL, and so my upstream bandwidth is less than my downstream. To prevent BT from consuming all my upstream bandwidth I am using the tc module in the kernel to restrict the BT packets (the rate limiting in BT is next to useless, as each instance of the client will use the programmed bandwidth - there is no "global" sharing of the bandwidth, so if you have 4 clients running it will take 4 times the bandwidth of 1 client).
The problem is that if the client is blocked sending an outbound torrent packet (because the traffic shaper queue is full), the client will not process any available incoming data packets, and this will hammer the download speed - I have expermimentally verified this.
Now, if there were separate threads for downloading and for uploading, the uploading threads would block as the TC queue filled, but the download threads would not be blocked, and could handle the download at full speed.
"thus simulating full duplex"
If there are separate channels for uplink and downlink, this is not simulating full duplex, it IS full duplex.
However, this makes me wonder about something. Compare this with another wireless mesh network - AX.25 (amatuer radio packet). In AX.25, rather than an end-to-end ack, each node acks the packet upon receipt, then forwards the packet to the next hop - thus acks need only move one hop.
Now, while this does increase the complexity of each node's processing (for example, how do you handle it when a packet has been accepted by a node, but that node cannot deliver the packet to the next hope), it cuts the delay on moving freight.
Now, TCP has a means to work around this (the TCP ack window), but it makes me wonder - would it be beneficial to a node system like this to have a protocol-layer store and forward system?
For example, what if each node ran a HTTP cache, and when a client requests a page, each node in the chain from client to server buffered the data so that any drop-outs and/or turnaround delays would have a minimal effect on the transfer?
I wonder which would have the greater effect upon throughput - full duplex, or caching? And would both have an even greater effect?
No, it should be called a "Mogolian" cluster.
Or "Chinese" cluster.
Becoming a republic won't help rid you of your royal family.
Just look at us: we've never had a royal family in our whole history as an independant country, yet we still have the Kennedys.
Or, now that I think about it, a flying game wherein the tilt of the laptop controls the yoke of the aircraft - imagine the fun of sitting in a cybercafe holding your Powerbook in your hands and going "Vrroommm! PowPowPow!"
Or rather, the fun of watching somebody else doing that....
Obviously, you all have no imagination.
What is clearly needed is a plug-in that clears the screen when the unit is held upside-down and shaken!
Or a version of Marble Madness that uses the tilt of the machine to control the marble.
You should quit when you have your next job lined up.
...." - those words WILL come back to haunt you (like, the *next* time you go to look for a new job, and prospective employers are calling this guy!) Make sure you give them your two weeks (they may offer to let you go immediately or ask that you continue to work - be ready either way).
That is, when you have the offer of employment from your new employer, and a starting date set.
I had a friend who did the "take this job and shove it" trick with what was truely a bad situation. However, it was several months before he had another job lined up, and he very nearly had to file for bankrupcy. It *did* screw his credit up for a long time, due to the amount of debt he racked up during that time.
All jobs suck - but some more than others.
So you should ask yourself, "Realistically, does this job suck worse than any new job I might get?"
Assuming the answer is "HELL YES!", then start looking for a new job - BUT DON'T LET YOUR CURRENT EMPLOYER KNOW. Make sure you tell any headhunters you work with that you don't want your current employer contacted.
Look long and well - do everything you can do to insure that your new job will suck less than your current job.
Then, when they offer you a position, set your start date no earlier than two and a half weeks into the future, get the formal (and legally binding) letter of offer and your letter of acceptance.
THEN, and ONLY then, do you go to your current boss and tender your resignation. And no matter how strong the temptation, no matter what you feel your justification is, no matter how badly you'd like to tell them off, resign in a calm, professional manner. This world is too damn small to say "First of all, you ain't no good, never been no good, you smell like old wet cheese, you pay shit
Also, when changing jobs, you are shaking your world up - so do your best to save up some emergency money before hand, and even if your new job pays 4x what you were making - act as though you were making your old salary and save the difference - at least for a year. Remember, last in, first out.
You may want to quit today - you may go home every night grinding your teeth, but USE that anger to drive your job search - remember, while your current job may suck, imagine how much MORE it will suck if you have to go crawling back in order to keep a roof over your head!
Craiglist: the 1990's called, and they want their dot-bomb business practices back.
Seriously - when I hear of a company doing dain-bramaged things like this, I think to myself "Well, there's a company with more money than common sense (or business sense)."
And of course, companies with more money than sense spontaniously decay, via moron emission, into companies with no cash and high debt, which then decay, via bankrupcy, into crap at auction.
Thanks for the catch - I do wish Mozilla would get a spell checker for edit dialogs within a web submit form.
Though that would not help the losing/loosing issue....
Don't take the absence of any moral obligation as prohibition.
In other words, just because a game designer/movie director/author/actor/... is not obligated to place any moral values into their story does not mean they should be prohibited from doing so if they so choose.
For example - I like the Myth series by Robert Lynn Asprin, precisely because RLA does put good "life lessions" into the stories.
Too many slashbots will jump on this story saying "Yeah, those bastards ought not be allowed to put their morality into their work!"
When the real point is "They ought not be REQUIRED to put SOMEBODY ELSE'S morality into their work."
You didn't actually FOLLOW THE LINK, did you?
I am not speaking of hunting Vulpes vulpes, but rather the sport practiced by amatuer radio operators of tracking a hidden transmitter with radio direction finding equipment.
This is just Foxhunting by another name.
The only difference is that this is open to non-hams.
The proper place for information like language preference is not on the card, but rather in the bank's database that the ATM accesses.
Ideally, when the card is first inserted the ATM will ask for non-secure data from the bank - things like language pref and such. If the card is NOT valid, the bank could send back default data (to prevent using that to ease checking of forged cards).
By seperating the prefs from the card, you can update the card without losing the prefs.
(Slashbots: Notice that the word is losing, not loosing!)