"We're not litigating. If we wanted to we would have done so years ago," said Horacio Gutierrez, Microsoft's VP for intellectual property and licensing, in an interview.
Are patents like trademarks, where if you don't actively enforce it you lose it? If they know of violations, but don't point it out to the violators and let it go on for years, do they lose their right to enforce the patent (at least for the violators they knew about)?
Threads are more efficient than processes for context switching because you don't have to switch all of the memory management state. It's been a while since I worked on Unix kernel context switch code (back when the 486 was common) but back then switching to a different process' page tables meant flushing the processor's paging caches.
Switching threads only requires saving the registers of one thread and restoring the registers of another.
Big difference.
Somebody educate me if this is more efficient in modern processors.
A whole host of options for protecting your ears. I have some of the Walker's Game Ears to protect me during shooting, hunting, and other noisy activities. The clip noises at 110db if I remember correctly.
I use Etymotic ear buds around the office to give a 25dB reduction in ambient chatter while mixing music with the various puter noises.
It's always nice to have someone suggest you're wrong, rather than inquire as to the differences between what they're doing and what you're doing which might account for us having different experiences.
It did take that long. I tried to just emerge the packages that were in my old world file, but got into all sorts of conflict issues. This seemed to have to do with virtual packages (e.g. mail agent) where I had chosen something other than the default. I kept getting complaints about it wanting to install conflicting packages, and had to piecemeal pick subsets of the packages from my old world file that didn't have conflicts. That certainly contributed to the elapsed time, since one emerge would complete sometime overnight and I'd start the next one in the morning when I woke up.
I also have a lot of software installed on this machine, since it's my primary development box: both gnome and kde, openoffice, evolution, firefox/thunderbird, gcc, subversion, cvs, several jdks, various video/audio tools, mail, dns, samba, dhcpd, vnc server and client, ldap, apache, tomcat, yada, yada, yada.
For what it's worth, it takes on the order of 12 hours to compile openoffice on my notebook. I have a p4 mobile, and only a 5400 rpm disk. Perhaps you're using a non-mobile processor or a faster disk.
This did all came about due to upgrading gcc/glibc to the latest stable version at the same time as an X.org upgrade. I had followed the instructions from the web site, including running revdep-rebuild, but things just got whacked.
Well, sort of. That'll get you the kernel sources, but you'll still have to compile them. You'll also have to generate a config to compile the kernel with. For that, you'll probably want to also emerge genkernel which is supposed to do a decent job of autogenerating the config for you. I haven't bothered with genkernel myself. I've been compiling kernels for so many years that it doesn't phase me to generate the config via the standard kernel make targets.
They have a binary install disk now that gets you a running system reasonably quickly, but whenever new versions of core packages are released (e.g. X.org, openoffice, evolution, gnome/kde) you'll spend a lot of time compiling the new stuff. If you want to change the compilation flags or package options (use flags), you'll need to recompile most if not everything on your system.
I'll say it again, in case someone thinks I'm bashing on Gentoo - I'm not. It's a perfectly fine distribution and has it's on pros like being able to use the exact compilation flags you want. But I have found over the past few years that it takes a fair bit of time and energy to keep things current and stable. I've just gotten to the point where I'd like to reduce the time I spend maintaining my server, and increase the time I spend actually doing something useful with it and having a life.
Um, let me get this straight. You're upset there is no 2.6.18 binary package included in Ubuntu which forces you to compile the kernel for yourself. In response, you're going to go to Gentoo and compile everything on your system for yourself. Do I have that right?
I'm actually getting ready to switch away from Gentoo. I've been using it for something like 4 or 5 years now, but I'm really getting tired of the compilation times and random changes that break my system and force me to hunt around the web site to find out which part of the system someone decided to rework. The only way I could get one of my systems back to running was to completely bebuild it, and it literally took me over a week of compiling on a 2GHz P4. That's no good when it's a notebook I use every day.
Don't take that as a complete condemnation of Gentoo; it just seems like keeping things running requires a lot more fiddling than Redhat did when I used it before. Now I'm trying to decide between Fedora Core vs. Ubuntu.
Yes and no. As an employee, it's unusual to get the preferred shares that the VCs and founders get. You get common stock instead. You don't control the company, the preferred share owners do. When a buyout comes along, the preferred share owners can decide to divvy up the money however they want. If they want to take all the money for themselves and leave nothing for the common stock owners, they can.
If they hire their buddy to work for the company, they can dilute your shares by issuing new ones to their friend. If they decide you're no longer pulling your weight, they can issue more to those who are and you again are diluted. Whenever there is a refinancing round, existing shares are diluted and those employees who are still with the company and still considered valuable are granted new shares (with a new vesting schedule) to partially undilute them. If a refinancing happens after you've left a private company, expect to get diluted.
VCs typically don't screw the employees by grabbing everything for themselves, but from what I've gathered from working in a variety of startups over the years they have the legal right to (IANAL, YMMV, yada, yada). It's bad for their rep if they screw employees over too badly, but they like to make sure they're in control if they're putting their money into the game.
I think the point was that once competition has been squashed they would be able to slack off and not worry about customer satisfaction.
I think the market will tend to self correct, but it takes longer to correct if there is an effective monopoly with no viable competition. Having a competitor in the ring forces a business to respond to customer needs more quickly.
At my previous company we built a system with on the order of 10 threads working on a combined dataset consisting of many hundreds of thousands of objects and occupying a couple hundred meg of memory in a large installation. There could be hundreds of thousands of instantiated locks in the system, although they fell into maybe 30 classes of lock. The large number of objects and locks was manageable because there were a small number of objects (say on the order of 25) that modeled something in the real world, but the system would then model a few hundred thousand of those real world things.
Anyway, suffice it to say it was a reasonably complicated realtime system with different threads having different responsibilities (receiving change data from the real world, work queues between different computation threads, pushing results back out to the real world).
To address the deadlock problem, we took the standard approach of defining an order in which locks must be taken to avoid deadlock. In our case, that really meant defining an ordering for the 30 or so distinct classes of locks and then defining a suborder for the few cases where we needed to lock multiple objects within a given class. Think of a class of lock/data as being a particular bit of subinformation for each of the real world things we were modeling.
Defining that order is hard, so I built a perl script that would extract partial orderings from comments in the code (e.g. "A : B C" would state that lock class A is taken while holding lock classes B and C) and then build a system wide ordering that obeys all the partial orderings. It's the same algorithm that make goes through to build things in the correct order.
Given that ordering, I then created a system to ensure we never took a lock out of order. First off, we wrapped the standard POSIX locking primitives with C++ classes, and created a smart lock handle we could instantiate on the stack and know that the lock would be released in its destructor no matter how you left the function (either through a return or an exception). We even went so far as to ensure that if you canceled a thread the locks would be released.
Given that wrapping, it was pretty straight forward to create a per-thread array to count how many locks were currently held at each level, and verify at lock time that no locks later in the hierarchy were currently held. In the event of an out of order lock attemt, it would throw a LockError exception. All of our exception classes included a stack trace, making it very easy to identify lock order violations just by looking at the logs.
Note that we weren't detecting actual deadlocks, we were detecting out of order locking that could potentially lead to deadlock at some point. This showed design errors very early on in development, rather than having the system deadlock on us at a customer site where things are difficult to debug.
I was going to open source the tools, but we were acquired by a large phone company and I ran head long into bureaucracy when I tried. I decided I'd rather just reimplement it all at some later point in time. I think I've given enough of a description here that you could do something similar if you want. I highly recommend it if you're building anything serious. It's very easy to accidentally take locks out of order if you're not careful, and that needs to be caught early on in development rather than at a customer.
Personally, my brand affinity for Tivo is higher than it is for DirecTV. I really like my DirecTivos but I wish DirecTV would let me have all the Tivo features. I'm staying with DirecTV for now because of inertia, but when I upgrade to HDTV I'll be dropping them in favor of a cablecard Tivo and downloading as much content as I can through Tivo's broadband service.
Everyone I know who has tried the home grown cable DVRs comments on how the interface sucks compared to Tivo.
Tivo may stay a niche / boutique player, but I think lots of us will stick with them for the quality of product they put out.
What's really scary is that Verizon seems to want a few dollars a month to allow you access to this "feature". They charge you $.10 per text message. They charge you a couple dollars to download a snippet of a song you already own to use as a ring tone.
They make tons of money for charging you for each thing you do, rather than just charging you for data access.
This is where the wired ISPs want to go as well. Rather than just charging you for data access to the internet, they want to figure out how much you'll pay depending on what the bits are. That'll be $.01 per instant message, $.10 per email, $.02/minute for VoIP, $1.00 to download/stream a movie, etc.
Why is it TiVO's web site doesn't just recognize mobile phone browsers as low bandwidth devices and give an light weight web page to do the same thing?
I did buy the "in home service" aimed at a home user. They came to my office because that's where I am during business hours.
I forgot a 3rd problem I had, which was a failed power supply. They overnighted me a new one, and I sent the bad one back in the same box. There was no cost to me.
One thing to note is that I had to pay extra for the in home service, but it was definitely worth it since I use my personal machine both all day at the office and then during the evenings when I get home.
I'm not sure if they do in home software service; I suspect that has to be done over the phone. But once something is identified as a hardware issue, Dell has been great.
Anyway, different people have good and bad experiences with different companies. The clincher for me is the ability to have a service tech come to me to fix hardware when it fails. I consider that a requirement for a notebook.
Sorry you had bad service at Dell. I've actually had the opposite experience, although it's largely with work that can't be outsourced to some place far away (i.e. requires physical access to the machine). That makes this post a bit off topic to the original thread of outsourced IT, but what what the heck.
I have an Inspiron 8500 and there was an initial fit and finish problem with the case that caused the display to turn off once the Dell video driver was loaded. I spent some time on the phone with them and the guy walked me through popping it open enough to fix the problem.
After a year or so, I also had a problem with my keyboard. I called in, did a quick verification that the problem still existed when in the BIOS screens (I run Linux, so the Windows based diagnostics couldn't be run) and they dispatched a service tech to my business to fix it the next day. The guy arrived, I came up front to meet him, and within half an hour I had a new keyboard installed in my machine. He also cleaned out enough cat hair to clothe a family of six, but that's another story.
I'm definitely very happy with Dell's in home service.
HP, on the other hand, was a much worse experience with my wife's notebook about a month and a half after purchase when the video went bad. HP doesn't offer any sort of in home service, so we had to Fedex the unit back to them. We also had to remove the hard drive before shipping because HP says they will always reimage your hard drive as part of servicing the computer. I'm sure that's due to all the problems they've had with Windows getting messed up and virus/spyware messing things around.
On top of that, instead of the quick turn around promised when we bought the notebook, they told us that since we had custom configured the unit instead of choosing one of their default configurations they had no obligation to fix it quickly. I think it was close to two weeks before we got it back.
After getting the unit back from HP, the wife is complaining that it is running slower than before. I have no idea how to address that, other than to verify it's the same CPU speed and memory configuration.
The moral of the story is to never buy a notebook that doesn't have the option of next day in home service. We will not buy from HP next time around, but we will buy from Dell.
This just seems like a really bad marketing move to me. Are there really that many people who would want a tablet PC built in to their coffee table? It seems analogous to trying to sell a desk with a built in PC. The whole point of the tablet and notebook market is that people like the portability market. Desktop machines are shrinking in size all the time, and the trend is away from desktops and towards notebooks. This is a step in the opposite direction of the industry trend.
I think they'd be far better off selling a normal tablet.
It would be very interesting to see the search engine and content providers band together and declare that any ISP trying to extort money in this way would have their address space filtered to cut their customers off from the combined pool of services and content. That would pretty much make it suicide for any ISP to go down that road.
It's probably considered illegal as anticompetitive behaviour. Would anyone who's not a lawyer but plays one on the internet care to offer insight into the legality of that approach?
If you don't mind spending money, get Rational Purify from IBM. They seem to properly support Linux now, and they are the gold standard. You can download a 2 week evaluation license to give it a try.
As far as your build process, you just take whatever your link command is, and tag "purify" on the beginning of it. The tool will instrument all compiled objects that you are linking in, and produce a binary that does all sorts of really useful checking (read/write of freed memory, buffer overruns, memory leaks, file descriptor leaks,...).
If you are developing software for money, there is nothing better.
If you're doing open source work for no pay, obviously it's hard to justify it.
Congresspeople get a lot of letters every week, and it would take a lot of time and money to write an individual response to each one of them. You indicate the bill number you are writing about, and a general pro or con view of it. They send back a form letter so they can save their money for other things. If they pay someone to respond individually to every letter, they would either have to increase their fund raising efforts or cut back on other things they do. Would you prefer either of those?
If you want better access, you could consider offering to donate time to your local political office, and field letters for them on technical areas. You would have to get over the fact that you would be required to respond with the elected official's viewpoints rather than your own, but it would allow you to potentially help influence the congressperson's views on the topic.
Also, a single letter on a particular subject is not important to them, nor should it be. They represent their entire constituency, and what matters is when they get many letters on the same topic. I believe they do listen when there is significant feedback. They tally the pro and con feedback, and try to get a sense of the public mood. I'm sure they also save all the letters, and eventually a staff member will scan over them and provide a summary of the points raised and perhaps a count of how many times each point was made by a different person. This information will then be used to help form the official position.
This brings up another difficulty in responding individually to each letter. An elected official doesn't know which way their constituents want them to lean on a particular bill until they've had a chance to go over all the feedback they've received on it. Given that, I think it's entirely appropriate to respond with a form letter indicating some of the main issues and stressing what their high level goals and beliefs are.
Also, you get to see the breakroom. Is it clean, spacious, stocked with food/drink or not? I've found that the breakroom is a great glimpse into the soul of an employer and a good way to see how you will be treated as an employee. If the breakroom is nothing but a sink with a giant poster stating "DO YOUR DISHES, I AM NOT YOUR MOMMY" that may be a hint that management is less than warm. Trust me on this one, I know.
Everywhere I've worked, it's not management who does the dishes in the break room. It's someone who is the office manager, an admin, or something like that. someone in one of these roles is usually catching all the little things that fall through the cracks and keeping the office running smoothly. So remember this, and do your damn dishes.:-)
It's not so much that they will select specific things to block, but rather they'll select specific things to be given preferred access.
In the router world, this is referred to as ToS (Type of Service) or QoS (Quality of Service). They are slightly different, but for the purpose of this conversation let's just say there is a single byte in every IP header that can be used to differentiate different kinds of traffic.
Routers also have the ability to have multiple outbound queues on a single hardware interface. You can configure a priority queue such that its packets are sent before any packets in the non-priority queue.
Force the byte to be 0 for any traffic coming in from a customer's site so they can't declare any traffic to be priority.
Set up an access control list (ACL) that matches traffic going to or from the service provider's audio server, video on demand server, etc. and sets the byte to 1.
Throughout your network, you configure priority queues that ensure your priority traffic gets transmitted first.
Given all this, the ISP can reduce the bandwidth of their backbone (or avoid increasing it as demand grows) and their pay-for-content services will work just fine but anyone else's services will suck.
The ISP can then go after other companies that are trying to sell content to their users. If Apple wishes to have priority access to the ISP's customers, they must pay a fee to have an ACL set up which flags their traffic as priority. Ditto for anyone selling a real-time stock market feed, video-on-demand, etc.
The ISP can then also target you as a customer. If you want to be able to receive any of this priority content, you'll have to pay an additional monthly fee to do so.
Personally, I don't like the idea of being charged differently based on who I'm talking to. It's like the post office or Fedex charging you more for a letter you're sending to your attorney because they know that must be important, but less for your letter to your mother. It's like when a truck enters a toll highway, they look inside to see what is being moved. If it's just a moving van full of personal belongings, the fee is low. But if it's a load of consumer electronics headed for sale they'll charge a higher fee.
I'd rather see this be done based on the level of service you're requesting. If you want low jitter, low latency access to the network, it costs more per Mbit than it does for high jitter, high latency access. Whether you have a voice call to your grandmother or your attorney, it shouldn't matter. Whether you're viewing a movie from the ISP's server of HBO's server, it shouldn't matter.
Unfortunately, the ISPs want to go the way of the cellular providers, to maximize their profits by charging you additional fees for anything they can get away with.
Every time this subject comes up, I want to try a ballot measure here in California. The idea is to force accountability about where they got my data from. I should be able to go in and audit not only the information they have on me, but also where they got it from. Similarly, anyone who sends me direct mail must include a traceback of where they got my information. This likely would be something like "ChoicePoint - CitiBank" or "Choicepoint - {Citibank, AT&T}", although realistically it would be "ChoicePoint - {every compnay I do business with}".
The idea is that every piece of junk mail that gets sent to me should clearly list which companies that I do business with are turning around and selling my information. I suspect this would provide some back pressure on companies, not just through shame but also because the added expense of tracking all this information. This will hopefully change the economics of junk mail such that they're forced to either abandon the practice or more accurately focus their efforts. Either way, I'd get less junk mail.
The alternate thought is to embrace junk mail and do everything I can to receive as much of it as possible, then burn it as a way to heat my home.
The restaurant analogy isn't quite right. Nobody will let you just come in and eat a meal. Instead, there are be a few companies that have built their own transportation systems (roads, rail, airports, etc.) and the only way for you to get to a restaurant is via one or more of their transportation systems. What's more, they refuse to allow you to just go get something to eat. Instead, you have to pay a monthly fee for all you can eat access to 50 low end restaurants. If you prefer better food, you can buy access to one or more bundles of 10 higher end restaurants, but only if you also buy access to the 50 low end restaurants first. You can't just buy access to some nice Thai restaurants, you also have to buy access to a bunch of low end diners and fast food joints that you have no interest in eating at. You also have to buy all you can eat access to some really bizarre fringe restaurants that cater to small groups like those people who want to eat bugs. These restaurants generally have nobody in them, but the corporations argue that this is good, because if you didn't help pick up the cost of keeping these fringe restaurants open then they wouldn't be able to make enough money to stay in business.
The corporations also require that the restaurants put ads on their menus, dishes, walls, etc. as an additional revenue stream.Some of these corporations have been selling commodity access to their transportation systems for purposes other than food as a way to get additional revenue. Others have built general purpose transportation systems and sold access to them so you can go anywhere you wish (e.g. parks, libraries, etc.). As the quality of these systems improves, both consumers and the restaurants are starting to realize it would be cheaper to just use a commodity transportation system to connect. The corporations realize that as well, and in an effort to maintain their profit margins are starting to threaten to introduce a toll fee if you're going to go directly to the restaurant yourself. It'll be a lower toll if you just want to go to the park, but if you want to go see a popular concert the toll will be high since you really want to go there.
It's like the phone company charging you a different rate depending on the content of your conversation. They'd charge you more to talk to your doctor than they would to talk to your mother.
I just don't believe this is the sort of model for the internet that people really want.
I think it's fine to differentiate based on the quality of the connection (e.g. low latency / jitter vs. high latency / jitter, low bandwidth vs. high bandwidth), but to differentiate on the type of data you're moving just seems wrong to me.
Yes, but isn't that the point of open source? I'm sure this homework report has been written by someone before, and if they'd just open source it then this guy could reuse it and save his time for writing some other more useful report that hasn't been written yet.
I've written assembly as well, but haven't needed to in years. And I'm not talking about a school project, I'm talking about real world code. Lots of it, over many, many years of my career.
Why on earth do you care about how clean the assembly language is for a particular chip? Do you care about how clean the microcode is in the computer in your car? Do you care about what goes on inside your TV or VCR? The vast majority of people buy a computer for the applications they can run on it, and never do any programming at all. Of the minority that do program, the vast majority are never exposed to anything but high level languages.
If you're writing the code generator for a compiler, I can see as you might care. But even there, there are larger issues. Apple is switching to Intel chips because Intel is achieving more performance for a given power/heat budget than the PPCs. Intel has economy of scale on its side as well. The end result is better for the user of the computer.
Sure, I agree that Intel's instruction set is not that pleasant to deal with, but seriously dude, get some perspective.
Slashdot Mirror
From TFA
"We're not litigating. If we wanted to we would have done so years ago," said Horacio Gutierrez, Microsoft's VP for intellectual property and licensing, in an interview.
Are patents like trademarks, where if you don't actively enforce it you lose it? If they know of violations, but don't point it out to the violators and let it go on for years, do they lose their right to enforce the patent (at least for the violators they knew about)?
Threads are more efficient than processes for context switching because you don't have to switch all of the memory management state. It's been a while since I worked on Unix kernel context switch code (back when the 486 was common) but back then switching to a different process' page tables meant flushing the processor's paging caches.
Switching threads only requires saving the registers of one thread and restoring the registers of another.
Big difference.
Somebody educate me if this is more efficient in modern processors.
A whole host of options for protecting your ears. I have some of the Walker's Game Ears to protect me during shooting, hunting, and other noisy activities. The clip noises at 110db if I remember correctly.
I use Etymotic ear buds around the office to give a 25dB reduction in ambient chatter while mixing music with the various puter noises.
It's always nice to have someone suggest you're wrong, rather than inquire as to the differences between what they're doing and what you're doing which might account for us having different experiences.
It did take that long. I tried to just emerge the packages that were in my old world file, but got into all sorts of conflict issues. This seemed to have to do with virtual packages (e.g. mail agent) where I had chosen something other than the default. I kept getting complaints about it wanting to install conflicting packages, and had to piecemeal pick subsets of the packages from my old world file that didn't have conflicts. That certainly contributed to the elapsed time, since one emerge would complete sometime overnight and I'd start the next one in the morning when I woke up.
I also have a lot of software installed on this machine, since it's my primary development box: both gnome and kde, openoffice, evolution, firefox/thunderbird, gcc, subversion, cvs, several jdks, various video/audio tools, mail, dns, samba, dhcpd, vnc server and client, ldap, apache, tomcat, yada, yada, yada.
For what it's worth, it takes on the order of 12 hours to compile openoffice on my notebook. I have a p4 mobile, and only a 5400 rpm disk. Perhaps you're using a non-mobile processor or a faster disk.
This did all came about due to upgrading gcc/glibc to the latest stable version at the same time as an X.org upgrade. I had followed the instructions from the web site, including running revdep-rebuild, but things just got whacked.
Well, sort of. That'll get you the kernel sources, but you'll still have to compile them. You'll also have to generate a config to compile the kernel with. For that, you'll probably want to also emerge genkernel which is supposed to do a decent job of autogenerating the config for you. I haven't bothered with genkernel myself. I've been compiling kernels for so many years that it doesn't phase me to generate the config via the standard kernel make targets.
They have a binary install disk now that gets you a running system reasonably quickly, but whenever new versions of core packages are released (e.g. X.org, openoffice, evolution, gnome/kde) you'll spend a lot of time compiling the new stuff. If you want to change the compilation flags or package options (use flags), you'll need to recompile most if not everything on your system.
I'll say it again, in case someone thinks I'm bashing on Gentoo - I'm not. It's a perfectly fine distribution and has it's on pros like being able to use the exact compilation flags you want. But I have found over the past few years that it takes a fair bit of time and energy to keep things current and stable. I've just gotten to the point where I'd like to reduce the time I spend maintaining my server, and increase the time I spend actually doing something useful with it and having a life.
Um, let me get this straight. You're upset there is no 2.6.18 binary package included in Ubuntu which forces you to compile the kernel for yourself. In response, you're going to go to Gentoo and compile everything on your system for yourself. Do I have that right?
I'm actually getting ready to switch away from Gentoo. I've been using it for something like 4 or 5 years now, but I'm really getting tired of the compilation times and random changes that break my system and force me to hunt around the web site to find out which part of the system someone decided to rework. The only way I could get one of my systems back to running was to completely bebuild it, and it literally took me over a week of compiling on a 2GHz P4. That's no good when it's a notebook I use every day.
Don't take that as a complete condemnation of Gentoo; it just seems like keeping things running requires a lot more fiddling than Redhat did when I used it before. Now I'm trying to decide between Fedora Core vs. Ubuntu.
Yes and no. As an employee, it's unusual to get the preferred shares that the VCs and founders get. You get common stock instead. You don't control the company, the preferred share owners do. When a buyout comes along, the preferred share owners can decide to divvy up the money however they want. If they want to take all the money for themselves and leave nothing for the common stock owners, they can.
If they hire their buddy to work for the company, they can dilute your shares by issuing new ones to their friend. If they decide you're no longer pulling your weight, they can issue more to those who are and you again are diluted. Whenever there is a refinancing round, existing shares are diluted and those employees who are still with the company and still considered valuable are granted new shares (with a new vesting schedule) to partially undilute them. If a refinancing happens after you've left a private company, expect to get diluted.
VCs typically don't screw the employees by grabbing everything for themselves, but from what I've gathered from working in a variety of startups over the years they have the legal right to (IANAL, YMMV, yada, yada). It's bad for their rep if they screw employees over too badly, but they like to make sure they're in control if they're putting their money into the game.
I think the point was that once competition has been squashed they would be able to slack off and not worry about customer satisfaction.
I think the market will tend to self correct, but it takes longer to correct if there is an effective monopoly with no viable competition. Having a competitor in the ring forces a business to respond to customer needs more quickly.
At my previous company we built a system with on the order of 10 threads working on a combined dataset consisting of many hundreds of thousands of objects and occupying a couple hundred meg of memory in a large installation. There could be hundreds of thousands of instantiated locks in the system, although they fell into maybe 30 classes of lock. The large number of objects and locks was manageable because there were a small number of objects (say on the order of 25) that modeled something in the real world, but the system would then model a few hundred thousand of those real world things.
Anyway, suffice it to say it was a reasonably complicated realtime system with different threads having different responsibilities (receiving change data from the real world, work queues between different computation threads, pushing results back out to the real world).
To address the deadlock problem, we took the standard approach of defining an order in which locks must be taken to avoid deadlock. In our case, that really meant defining an ordering for the 30 or so distinct classes of locks and then defining a suborder for the few cases where we needed to lock multiple objects within a given class. Think of a class of lock/data as being a particular bit of subinformation for each of the real world things we were modeling.
Defining that order is hard, so I built a perl script that would extract partial orderings from comments in the code (e.g. "A : B C" would state that lock class A is taken while holding lock classes B and C) and then build a system wide ordering that obeys all the partial orderings. It's the same algorithm that make goes through to build things in the correct order.
Given that ordering, I then created a system to ensure we never took a lock out of order. First off, we wrapped the standard POSIX locking primitives with C++ classes, and created a smart lock handle we could instantiate on the stack and know that the lock would be released in its destructor no matter how you left the function (either through a return or an exception). We even went so far as to ensure that if you canceled a thread the locks would be released.
Given that wrapping, it was pretty straight forward to create a per-thread array to count how many locks were currently held at each level, and verify at lock time that no locks later in the hierarchy were currently held. In the event of an out of order lock attemt, it would throw a LockError exception. All of our exception classes included a stack trace, making it very easy to identify lock order violations just by looking at the logs.
Note that we weren't detecting actual deadlocks, we were detecting out of order locking that could potentially lead to deadlock at some point. This showed design errors very early on in development, rather than having the system deadlock on us at a customer site where things are difficult to debug.
I was going to open source the tools, but we were acquired by a large phone company and I ran head long into bureaucracy when I tried. I decided I'd rather just reimplement it all at some later point in time. I think I've given enough of a description here that you could do something similar if you want. I highly recommend it if you're building anything serious. It's very easy to accidentally take locks out of order if you're not careful, and that needs to be caught early on in development rather than at a customer.
Personally, my brand affinity for Tivo is higher than it is for DirecTV. I really like my DirecTivos but I wish DirecTV would let me have all the Tivo features. I'm staying with DirecTV for now because of inertia, but when I upgrade to HDTV I'll be dropping them in favor of a cablecard Tivo and downloading as much content as I can through Tivo's broadband service.
Everyone I know who has tried the home grown cable DVRs comments on how the interface sucks compared to Tivo.
Tivo may stay a niche / boutique player, but I think lots of us will stick with them for the quality of product they put out.
What's really scary is that Verizon seems to want a few dollars a month to allow you access to this "feature". They charge you $.10 per text message. They charge you a couple dollars to download a snippet of a song you already own to use as a ring tone.
They make tons of money for charging you for each thing you do, rather than just charging you for data access.
This is where the wired ISPs want to go as well. Rather than just charging you for data access to the internet, they want to figure out how much you'll pay depending on what the bits are. That'll be $.01 per instant message, $.10 per email, $.02/minute for VoIP, $1.00 to download/stream a movie, etc.
Why is it TiVO's web site doesn't just recognize mobile phone browsers as low bandwidth devices and give an light weight web page to do the same thing?
I did buy the "in home service" aimed at a home user. They came to my office because that's where I am during business hours.
I forgot a 3rd problem I had, which was a failed power supply. They overnighted me a new one, and I sent the bad one back in the same box. There was no cost to me.
One thing to note is that I had to pay extra for the in home service, but it was definitely worth it since I use my personal machine both all day at the office and then during the evenings when I get home.
I'm not sure if they do in home software service; I suspect that has to be done over the phone. But once something is identified as a hardware issue, Dell has been great.
Anyway, different people have good and bad experiences with different companies. The clincher for me is the ability to have a service tech come to me to fix hardware when it fails. I consider that a requirement for a notebook.
Cheers.
Sorry you had bad service at Dell. I've actually had the opposite experience, although it's largely with work that can't be outsourced to some place far away (i.e. requires physical access to the machine). That makes this post a bit off topic to the original thread of outsourced IT, but what what the heck.
I have an Inspiron 8500 and there was an initial fit and finish problem with the case that caused the display to turn off once the Dell video driver was loaded. I spent some time on the phone with them and the guy walked me through popping it open enough to fix the problem.
After a year or so, I also had a problem with my keyboard. I called in, did a quick verification that the problem still existed when in the BIOS screens (I run Linux, so the Windows based diagnostics couldn't be run) and they dispatched a service tech to my business to fix it the next day. The guy arrived, I came up front to meet him, and within half an hour I had a new keyboard installed in my machine. He also cleaned out enough cat hair to clothe a family of six, but that's another story.
I'm definitely very happy with Dell's in home service.
HP, on the other hand, was a much worse experience with my wife's notebook about a month and a half after purchase when the video went bad. HP doesn't offer any sort of in home service, so we had to Fedex the unit back to them. We also had to remove the hard drive before shipping because HP says they will always reimage your hard drive as part of servicing the computer. I'm sure that's due to all the problems they've had with Windows getting messed up and virus/spyware messing things around.
On top of that, instead of the quick turn around promised when we bought the notebook, they told us that since we had custom configured the unit instead of choosing one of their default configurations they had no obligation to fix it quickly. I think it was close to two weeks before we got it back.
After getting the unit back from HP, the wife is complaining that it is running slower than before. I have no idea how to address that, other than to verify it's the same CPU speed and memory configuration.
The moral of the story is to never buy a notebook that doesn't have the option of next day in home service. We will not buy from HP next time around, but we will buy from Dell.
This just seems like a really bad marketing move to me. Are there really that many people who would want a tablet PC built in to their coffee table? It seems analogous to trying to sell a desk with a built in PC. The whole point of the tablet and notebook market is that people like the portability market. Desktop machines are shrinking in size all the time, and the trend is away from desktops and towards notebooks. This is a step in the opposite direction of the industry trend.
I think they'd be far better off selling a normal tablet.
Agreed that it's pure nonsense.
It would be very interesting to see the search engine and content providers band together and declare that any ISP trying to extort money in this way would have their address space filtered to cut their customers off from the combined pool of services and content. That would pretty much make it suicide for any ISP to go down that road.
It's probably considered illegal as anticompetitive behaviour. Would anyone who's not a lawyer but plays one on the internet care to offer insight into the legality of that approach?
If you don't mind spending money, get Rational Purify from IBM. They seem to properly support Linux now, and they are the gold standard. You can download a 2 week evaluation license to give it a try.
As far as your build process, you just take whatever your link command is, and tag "purify" on the beginning of it. The tool will instrument all compiled objects that you are linking in, and produce a binary that does all sorts of really useful checking (read/write of freed memory, buffer overruns, memory leaks, file descriptor leaks, ...).
If you are developing software for money, there is nothing better.
If you're doing open source work for no pay, obviously it's hard to justify it.
Congresspeople get a lot of letters every week, and it would take a lot of time and money to write an individual response to each one of them. You indicate the bill number you are writing about, and a general pro or con view of it. They send back a form letter so they can save their money for other things. If they pay someone to respond individually to every letter, they would either have to increase their fund raising efforts or cut back on other things they do. Would you prefer either of those?
If you want better access, you could consider offering to donate time to your local political office, and field letters for them on technical areas. You would have to get over the fact that you would be required to respond with the elected official's viewpoints rather than your own, but it would allow you to potentially help influence the congressperson's views on the topic.
Also, a single letter on a particular subject is not important to them, nor should it be. They represent their entire constituency, and what matters is when they get many letters on the same topic. I believe they do listen when there is significant feedback. They tally the pro and con feedback, and try to get a sense of the public mood. I'm sure they also save all the letters, and eventually a staff member will scan over them and provide a summary of the points raised and perhaps a count of how many times each point was made by a different person. This information will then be used to help form the official position.
This brings up another difficulty in responding individually to each letter. An elected official doesn't know which way their constituents want them to lean on a particular bill until they've had a chance to go over all the feedback they've received on it. Given that, I think it's entirely appropriate to respond with a form letter indicating some of the main issues and stressing what their high level goals and beliefs are.
Everywhere I've worked, it's not management who does the dishes in the break room. It's someone who is the office manager, an admin, or something like that. someone in one of these roles is usually catching all the little things that fall through the cracks and keeping the office running smoothly. So remember this, and do your damn dishes. :-)
It's not so much that they will select specific things to block, but rather they'll select specific things to be given preferred access.
In the router world, this is referred to as ToS (Type of Service) or QoS (Quality of Service). They are slightly different, but for the purpose of this conversation let's just say there is a single byte in every IP header that can be used to differentiate different kinds of traffic.
Routers also have the ability to have multiple outbound queues on a single hardware interface. You can configure a priority queue such that its packets are sent before any packets in the non-priority queue.
Given all this, the ISP can reduce the bandwidth of their backbone (or avoid increasing it as demand grows) and their pay-for-content services will work just fine but anyone else's services will suck.
The ISP can then go after other companies that are trying to sell content to their users. If Apple wishes to have priority access to the ISP's customers, they must pay a fee to have an ACL set up which flags their traffic as priority. Ditto for anyone selling a real-time stock market feed, video-on-demand, etc.
The ISP can then also target you as a customer. If you want to be able to receive any of this priority content, you'll have to pay an additional monthly fee to do so.
Personally, I don't like the idea of being charged differently based on who I'm talking to. It's like the post office or Fedex charging you more for a letter you're sending to your attorney because they know that must be important, but less for your letter to your mother. It's like when a truck enters a toll highway, they look inside to see what is being moved. If it's just a moving van full of personal belongings, the fee is low. But if it's a load of consumer electronics headed for sale they'll charge a higher fee.
I'd rather see this be done based on the level of service you're requesting. If you want low jitter, low latency access to the network, it costs more per Mbit than it does for high jitter, high latency access. Whether you have a voice call to your grandmother or your attorney, it shouldn't matter. Whether you're viewing a movie from the ISP's server of HBO's server, it shouldn't matter.
Unfortunately, the ISPs want to go the way of the cellular providers, to maximize their profits by charging you additional fees for anything they can get away with.
Where do you live, and is the lot next door for sale?
Every time this subject comes up, I want to try a ballot measure here in California. The idea is to force accountability about where they got my data from. I should be able to go in and audit not only the information they have on me, but also where they got it from. Similarly, anyone who sends me direct mail must include a traceback of where they got my information. This likely would be something like "ChoicePoint - CitiBank" or "Choicepoint - {Citibank, AT&T}", although realistically it would be "ChoicePoint - {every compnay I do business with}".
The idea is that every piece of junk mail that gets sent to me should clearly list which companies that I do business with are turning around and selling my information. I suspect this would provide some back pressure on companies, not just through shame but also because the added expense of tracking all this information. This will hopefully change the economics of junk mail such that they're forced to either abandon the practice or more accurately focus their efforts. Either way, I'd get less junk mail.
The alternate thought is to embrace junk mail and do everything I can to receive as much of it as possible, then burn it as a way to heat my home.
The restaurant analogy isn't quite right. Nobody will let you just come in and eat a meal. Instead, there are be a few companies that have built their own transportation systems (roads, rail, airports, etc.) and the only way for you to get to a restaurant is via one or more of their transportation systems. What's more, they refuse to allow you to just go get something to eat. Instead, you have to pay a monthly fee for all you can eat access to 50 low end restaurants. If you prefer better food, you can buy access to one or more bundles of 10 higher end restaurants, but only if you also buy access to the 50 low end restaurants first. You can't just buy access to some nice Thai restaurants, you also have to buy access to a bunch of low end diners and fast food joints that you have no interest in eating at. You also have to buy all you can eat access to some really bizarre fringe restaurants that cater to small groups like those people who want to eat bugs. These restaurants generally have nobody in them, but the corporations argue that this is good, because if you didn't help pick up the cost of keeping these fringe restaurants open then they wouldn't be able to make enough money to stay in business.
The corporations also require that the restaurants put ads on their menus, dishes, walls, etc. as an additional revenue stream.Some of these corporations have been selling commodity access to their transportation systems for purposes other than food as a way to get additional revenue. Others have built general purpose transportation systems and sold access to them so you can go anywhere you wish (e.g. parks, libraries, etc.). As the quality of these systems improves, both consumers and the restaurants are starting to realize it would be cheaper to just use a commodity transportation system to connect. The corporations realize that as well, and in an effort to maintain their profit margins are starting to threaten to introduce a toll fee if you're going to go directly to the restaurant yourself. It'll be a lower toll if you just want to go to the park, but if you want to go see a popular concert the toll will be high since you really want to go there.
It's like the phone company charging you a different rate depending on the content of your conversation. They'd charge you more to talk to your doctor than they would to talk to your mother.
I just don't believe this is the sort of model for the internet that people really want.
I think it's fine to differentiate based on the quality of the connection (e.g. low latency / jitter vs. high latency / jitter, low bandwidth vs. high bandwidth), but to differentiate on the type of data you're moving just seems wrong to me.
Yes, but isn't that the point of open source? I'm sure this homework report has been written by someone before, and if they'd just open source it then this guy could reuse it and save his time for writing some other more useful report that hasn't been written yet.
Snicker.
I've written assembly as well, but haven't needed to in years. And I'm not talking about a school project, I'm talking about real world code. Lots of it, over many, many years of my career.
Why on earth do you care about how clean the assembly language is for a particular chip? Do you care about how clean the microcode is in the computer in your car? Do you care about what goes on inside your TV or VCR? The vast majority of people buy a computer for the applications they can run on it, and never do any programming at all. Of the minority that do program, the vast majority are never exposed to anything but high level languages.
If you're writing the code generator for a compiler, I can see as you might care. But even there, there are larger issues. Apple is switching to Intel chips because Intel is achieving more performance for a given power/heat budget than the PPCs. Intel has economy of scale on its side as well. The end result is better for the user of the computer.
Sure, I agree that Intel's instruction set is not that pleasant to deal with, but seriously dude, get some perspective.
People who pay SCO their extortion fees suck cock by choice.