At work we have some 3rd-party web service for our Customer relationship management system and it uses a lot of scripts and objects. Chrome hesitates a bit, but while some of the objects are loading, I can change tabs. With Firefox, not only does the current tab completely freeze during rendering, but I can't change tabs until it finishes.
Agile is a programming methodology, not an architect or design methodology. Agile is meant to allow you to better handle feature creep with less pain because you can never fully get rid of feature creep. But if you skip design and jump strait into programming, all features are feature creep and you're going to have a bad time.
I was born into a poor family, couldn't even afford a TV until I was in my teens in the mid 90s, and I was instantly made middle class shortly after graduating from my nearly free state Uni during the recent recession. I had my pick of the litter when it came to jobs, I was flooded with companies wanting immediately to hire me into the top 20% of local incomes. Yay for programming.
I ended up holding off and turning down several jobs in 2008 because I really wanted a local job near family. The best part is I didn't need to do any job hunting, they came to me. Calling my phone, sending me mail.
You should try working on the Linux kernel. Linus has a much shorter fuse for idiots. Sometimes it's better to scare some people off than to let them poison the code with their ineptitude. FreeBSD is a meritocracy. Quality over political correctness.
Anyone who has worked on infrastructure projects will understand the necessity of a good design's importance to long term efficiency. Linus has made sure that the Linux kernel is a stable dependency on which to build your projects, but userland has no such thing. Everything is in constant flux. "Not invented here" is a rampant issue. Fix you current stuff before reinventing a slightly different wheel.
New does not mean buggy unless you have piss poor programmers.
I didn't live through it, but based on what I've read and watched about the evolution of FreeBSD, they got set back for a long while from bogus legal issues and a resulting anemic community. It's well engineered base has allowed it to quickly catch up and is now starting to surpass Linux in *some* ways. While FreeBSD is lacking in some ways like a large driver portfolio and mind share of the popular crowd, the fundamental parts of the OS are starting to set the bar for high performance and stability. FreeBSD already has much superior documentation, but there is a new initiative to make it much better.
With a good foundation of performance, stability, consistency, documentation, near seamless integration of stuff like ZFS and Jails, it is only a matter of time before it starts to get popular as an alternative. I hear PC-BSD 11 has made some huge changes to making the desktop experience even better than it already was.
If there is anything that I've learned as a programmer it's that when something is well designed, it can change to meet a moving target of expectations without major changes. The Linux community has a whole has some ADD notion that tools are disposable and to replace them with the latest greatest tool. This is just a sign that no one put any thought into the original tool.
FreeBSD implemented Jails back in 2000. Linux got Linux Containers in 2008, but still not nearly as good as Jails. Now Docker is gaining transaction. They really are all the same things. FreeBSD is getting work done to emulate Docker with Jails because Docker is only handles a subset of what Jails can do. Do it once, do it right, stop creating so many half-assed versions of the same thing!
FreeBSD foundation has a server that was originally FreeBSD 5.0 32bit and has been in-place upgraded all the way up to FreeBSD 9.3 64bit, not to mention migrated through several physical servers through the years. There are still original FreeBSD 5.0 binaries that are still running and to which they no longer have the original source code.
Even when pushed beyond any sane limits, FreeBSD keeps on trucking.
110'F is very rare. Rarely even hit 100f in a given year. -40f isn't too far off, more likely than 100f. All the really bad winter storms are weakened by the time they pass over Minn.
The issue is peers with massive bufferbloat and using TCP. I do a lot of seeding because of my symmetrical connection so I have my connection limit up quite high. Normally I have a solid 9ms ping to Chicago, but when I'm getting DDOS'd by a bunch of bufferbloated seeders to the point that I am getting 20% packetloss, my ping is at a sky-high 45ms! That's also some of the issue. Even under a DDOS my ping never goes higher than about 35ms over idle.
While I got TCP on bufferbloated connections sending 20Mb/s+ of duplicate packets, uTP isn't backing off correctly because its primary metric for measuring congestion is latency. My latency at max only increase by 35ms, even with 20% loss from the DDOS. No bufferbloat on my ISP here. Strange things happens when congestion control algorithms assume latency goes up dramatically with congestion.
The saving grace is qBitTorrent was self limited to 10MiB/s and I was getting exactly that, even with all the loss. I think my ISP uses a fair queuing AQM, so ICMP may be seeing loss because it just so happened to share a bucket with a heavy flow, but any individual data flow may not. This allows some subset of my connections to run mostly unaffected.
Holy crap bufferbloat! 20% packetloss. No, not my connection, everyone else. Egress on my LAN is only 80Mb/s, but ingress on my WAN is over 100Mb/s and maxing my connection. Trace route some of those peers that I was downloading from. Low pings all the way into their ISP, then 1-2 hops before hitting the peer, pings skyrocket into the thousands.
Why am I seeing a 20Mb/s difference between WAN ingress and LAN egress? Those are all retransmit packets that my stateful firewall is filtering out.
If your connection has massive amounts of bufferbloat, please don't seed. You're DOS'n my connection.
I get all over-the-air TV channels, plus extra for $15/m plus taxes and fees, which works out to just under $20. HD CBS, HD ABC, HD FOX, HD PBS, HD NBC, PBS 1, PBS 2, PBS 3, HD TNT, HD USA, HD ABC Family, C-SPAN, 5 different community access channels, free VoD. All of the HD channels get 24 hour time shifting for free, and of course paid VoD.
The wife has fallen in love with 24 hour time shifting. No more commercials. Can even watch shows the next day because you know, "24 hours". She's no longer tied to being at home at certain times.
Yeah, the government shouldn't force anyone to do anything, but there is some truth to the notion of better support. I'm not going to use Microsoft because they're "decent", but I will use cheap firewall/router appliances. We need lemon laws. If something remains unsecure with known attacks for X amount of time and you've purchased it in the last 5 years, you should be able to return for a full refund, free S&H.
Vulkan is based on Mantle. They whole-sale lifted nearly all of the API signatures and just renamed them. Kronos explicitly stated that they do not create new APIs, the only define established ones. They are followers, not leaders.
Vulkan shouldn't need to be extensible. I'm sure there will be a few exceptions, but I don't expect nearly as many. The old API needed to do most of the work, so implementation was important. Vulkan now expect the game engine to do the implementation and to communicate near-directly with the GPU. Extending Vulkan would be akin to extending x86 to make your program work. Someone may want to may a new SIMD, but it should be very rare to need a new extension.
The whole draw of OpenGL was that developers could just add in the effects they wanted
It is also its greatest weakness. There are two independent sets of developers, the graphic driver developers and the developers who actually use the API. The graphic developers may love that they can extend OpenGL to make use of their random new fangled features in their graphics card, but the developers actually using the API now need to check for some huge number of custom extensions that may or may not exist.
The whole point of a standard API is for there to be a standard! either you follow the standard or you don't and everyone who implements the standard implements the exact same things.
If you force someone to charge more for labour than that labour provides in value
Depends on your definition of "value". Most people conflate net income of money and value. An example of confusion is the government can give away a service, say costs $1000 and a pure monetary loss, but still add value. Money has no inherent value. Something of value increases productivity of man-hours more than the cost man-hours it took to create it. Something can have negative value such that it produces less productivity that the cost of creating it.
Transportation infrastructure is an example. It can cost billions and a private company may never make that money back, but if given away for free, can increase the size of the market by much more than its cost. Not all value can be represented with money.
An example of negative value would be someone abusing monopolistic power to price gouge. Money does represent value and charging more money for something than it is worth, but customers do not have a choice, reduces value because less money can be spent in areas that better deserve it.
What's the point of life if you can't slave at work?
Must fix Windows Metro Netflix app that keeps rebuffering video streams and maxing out my connection. Averaging 60Mb/s for 3 minutes because the app keeps forgetting the data it already downloaded. At one point I had a 30Mb/s average over an hour, that was just me watching on one device. Something is seriously wrong. Web app works just fine, but no SuperHD:*(
I forgot to also mention, the topic gets even more convoluted once you include that the eye amplifies contrast, which can increase human perception of color near contrasting edges.
Because the normal human cannot distinguish shades of color beyond 32 bit RGB.
Really? I see color banding all the time in games. When color banding is no longer an issue, then the color depth will be enough. A quick google returns some answers saying the usage of the term "color" is ambiguous in most cases or miss-understood. If you define "color" as how a layman would use, we can see closer to 100 million colors. Most usages of the term "color" does not include luminosity.
Slashdot Mirror
At work we have some 3rd-party web service for our Customer relationship management system and it uses a lot of scripts and objects. Chrome hesitates a bit, but while some of the objects are loading, I can change tabs. With Firefox, not only does the current tab completely freeze during rendering, but I can't change tabs until it finishes.
Agile is a programming methodology, not an architect or design methodology. Agile is meant to allow you to better handle feature creep with less pain because you can never fully get rid of feature creep. But if you skip design and jump strait into programming, all features are feature creep and you're going to have a bad time.
I was born into a poor family, couldn't even afford a TV until I was in my teens in the mid 90s, and I was instantly made middle class shortly after graduating from my nearly free state Uni during the recent recession. I had my pick of the litter when it came to jobs, I was flooded with companies wanting immediately to hire me into the top 20% of local incomes. Yay for programming.
I ended up holding off and turning down several jobs in 2008 because I really wanted a local job near family. The best part is I didn't need to do any job hunting, they came to me. Calling my phone, sending me mail.
You should try working on the Linux kernel. Linus has a much shorter fuse for idiots. Sometimes it's better to scare some people off than to let them poison the code with their ineptitude. FreeBSD is a meritocracy. Quality over political correctness.
Anyone who has worked on infrastructure projects will understand the necessity of a good design's importance to long term efficiency. Linus has made sure that the Linux kernel is a stable dependency on which to build your projects, but userland has no such thing. Everything is in constant flux. "Not invented here" is a rampant issue. Fix you current stuff before reinventing a slightly different wheel.
New does not mean buggy unless you have piss poor programmers.
I didn't live through it, but based on what I've read and watched about the evolution of FreeBSD, they got set back for a long while from bogus legal issues and a resulting anemic community. It's well engineered base has allowed it to quickly catch up and is now starting to surpass Linux in *some* ways. While FreeBSD is lacking in some ways like a large driver portfolio and mind share of the popular crowd, the fundamental parts of the OS are starting to set the bar for high performance and stability. FreeBSD already has much superior documentation, but there is a new initiative to make it much better.
With a good foundation of performance, stability, consistency, documentation, near seamless integration of stuff like ZFS and Jails, it is only a matter of time before it starts to get popular as an alternative. I hear PC-BSD 11 has made some huge changes to making the desktop experience even better than it already was.
If there is anything that I've learned as a programmer it's that when something is well designed, it can change to meet a moving target of expectations without major changes. The Linux community has a whole has some ADD notion that tools are disposable and to replace them with the latest greatest tool. This is just a sign that no one put any thought into the original tool.
FreeBSD implemented Jails back in 2000. Linux got Linux Containers in 2008, but still not nearly as good as Jails. Now Docker is gaining transaction. They really are all the same things. FreeBSD is getting work done to emulate Docker with Jails because Docker is only handles a subset of what Jails can do. Do it once, do it right, stop creating so many half-assed versions of the same thing!
FreeBSD foundation has a server that was originally FreeBSD 5.0 32bit and has been in-place upgraded all the way up to FreeBSD 9.3 64bit, not to mention migrated through several physical servers through the years. There are still original FreeBSD 5.0 binaries that are still running and to which they no longer have the original source code.
Even when pushed beyond any sane limits, FreeBSD keeps on trucking.
"employee benefits, fleet maintenance, etc" come out of revenue. All it means is Uber has more inflow of money, nothing to do about profits.
Chicago 9ms
:-)
New York City 35ms
Atlanta 40ms
Dallas Texas 40ms
Miami 45ms
LA 60ms
London 90ms
France 90ms
Frankfurt 110ms
Heck, New Zealand is only 200ms and under 5ms of jitter
Network Graphs: https://lh3.googleusercontent....
DSLReports Bufferbloat 32/16 Test: https://lh3.googleusercontent....
DSLReports Jitter Test: https://lh3.googleusercontent....
64KiB buffer. Divide by latency to get theoretical max throughput. 64KiB/0.1sec(100ms) = 614KiB/s. Yeah, pretty slow.
110'F is very rare. Rarely even hit 100f in a given year. -40f isn't too far off, more likely than 100f. All the really bad winter storms are weakened by the time they pass over Minn.
The issue is peers with massive bufferbloat and using TCP. I do a lot of seeding because of my symmetrical connection so I have my connection limit up quite high. Normally I have a solid 9ms ping to Chicago, but when I'm getting DDOS'd by a bunch of bufferbloated seeders to the point that I am getting 20% packetloss, my ping is at a sky-high 45ms! That's also some of the issue. Even under a DDOS my ping never goes higher than about 35ms over idle.
While I got TCP on bufferbloated connections sending 20Mb/s+ of duplicate packets, uTP isn't backing off correctly because its primary metric for measuring congestion is latency. My latency at max only increase by 35ms, even with 20% loss from the DDOS. No bufferbloat on my ISP here. Strange things happens when congestion control algorithms assume latency goes up dramatically with congestion.
The saving grace is qBitTorrent was self limited to 10MiB/s and I was getting exactly that, even with all the loss. I think my ISP uses a fair queuing AQM, so ICMP may be seeing loss because it just so happened to share a bucket with a heavy flow, but any individual data flow may not. This allows some subset of my connections to run mostly unaffected.
Holy crap bufferbloat! 20% packetloss. No, not my connection, everyone else. Egress on my LAN is only 80Mb/s, but ingress on my WAN is over 100Mb/s and maxing my connection. Trace route some of those peers that I was downloading from. Low pings all the way into their ISP, then 1-2 hops before hitting the peer, pings skyrocket into the thousands.
Why am I seeing a 20Mb/s difference between WAN ingress and LAN egress? Those are all retransmit packets that my stateful firewall is filtering out.
If your connection has massive amounts of bufferbloat, please don't seed. You're DOS'n my connection.
Coming out of their share of of the entertainment funds. They can have their $100/m of extra channels, I'm getting a new computer every year.
I get all over-the-air TV channels, plus extra for $15/m plus taxes and fees, which works out to just under $20. HD CBS, HD ABC, HD FOX, HD PBS, HD NBC, PBS 1, PBS 2, PBS 3, HD TNT, HD USA, HD ABC Family, C-SPAN, 5 different community access channels, free VoD. All of the HD channels get 24 hour time shifting for free, and of course paid VoD.
The wife has fallen in love with 24 hour time shifting. No more commercials. Can even watch shows the next day because you know, "24 hours". She's no longer tied to being at home at certain times.
with the Internet, who pays for porn?
http://arstechnica.com/busines...
Technically I have IPTV, but it requires a proprietary appliance. No more coax in the house, everything from my ISP is over Ethernet.
Yeah, the government shouldn't force anyone to do anything, but there is some truth to the notion of better support. I'm not going to use Microsoft because they're "decent", but I will use cheap firewall/router appliances. We need lemon laws. If something remains unsecure with known attacks for X amount of time and you've purchased it in the last 5 years, you should be able to return for a full refund, free S&H.
Vulkan is based on Mantle. They whole-sale lifted nearly all of the API signatures and just renamed them. Kronos explicitly stated that they do not create new APIs, the only define established ones. They are followers, not leaders.
Vulkan shouldn't need to be extensible. I'm sure there will be a few exceptions, but I don't expect nearly as many. The old API needed to do most of the work, so implementation was important. Vulkan now expect the game engine to do the implementation and to communicate near-directly with the GPU. Extending Vulkan would be akin to extending x86 to make your program work. Someone may want to may a new SIMD, but it should be very rare to need a new extension.
The whole draw of OpenGL was that developers could just add in the effects they wanted
It is also its greatest weakness. There are two independent sets of developers, the graphic driver developers and the developers who actually use the API. The graphic developers may love that they can extend OpenGL to make use of their random new fangled features in their graphics card, but the developers actually using the API now need to check for some huge number of custom extensions that may or may not exist.
The whole point of a standard API is for there to be a standard! either you follow the standard or you don't and everyone who implements the standard implements the exact same things.
If you force someone to charge more for labour than that labour provides in value
Depends on your definition of "value". Most people conflate net income of money and value. An example of confusion is the government can give away a service, say costs $1000 and a pure monetary loss, but still add value. Money has no inherent value. Something of value increases productivity of man-hours more than the cost man-hours it took to create it. Something can have negative value such that it produces less productivity that the cost of creating it.
Transportation infrastructure is an example. It can cost billions and a private company may never make that money back, but if given away for free, can increase the size of the market by much more than its cost. Not all value can be represented with money.
An example of negative value would be someone abusing monopolistic power to price gouge. Money does represent value and charging more money for something than it is worth, but customers do not have a choice, reduces value because less money can be spent in areas that better deserve it.
What's the point of life if you can't slave at work?
:*(
Must fix Windows Metro Netflix app that keeps rebuffering video streams and maxing out my connection. Averaging 60Mb/s for 3 minutes because the app keeps forgetting the data it already downloaded. At one point I had a 30Mb/s average over an hour, that was just me watching on one device. Something is seriously wrong. Web app works just fine, but no SuperHD
I forgot to also mention, the topic gets even more convoluted once you include that the eye amplifies contrast, which can increase human perception of color near contrasting edges.
Because the normal human cannot distinguish shades of color beyond 32 bit RGB.
Really? I see color banding all the time in games. When color banding is no longer an issue, then the color depth will be enough. A quick google returns some answers saying the usage of the term "color" is ambiguous in most cases or miss-understood. If you define "color" as how a layman would use, we can see closer to 100 million colors. Most usages of the term "color" does not include luminosity.