I'm a perfectionist, and I pretty much don't give up before things work. I moved to Windows on my laptopm after having tried to write a module for pulseaudio, which took me tens of hours and got me 1 % of the way to perfection. Why I left Ubuntu on the laptop:
Tearing artifacts in video. Except for once on the primary laptop display of a laptop, I haven't seen a single Linux or Unix install that does video playback without artifacts.
Sleep/hibernate. Hangs about every 20th suspend/resume cycle, and USB ports, USB hubs stop working more often than that.
Using external monitors: It didn't detect the external screen automatically on my laptop, and it didn't remember my preference.
Using different sound cards, such as a USB headset at work and a USB soundcard at home. It never changed over to those automatically. HDMI audio never worked, and if it was fixed, I'm sure it wouldn't switch automatically when I plugged in a TV
I kind of got a Dolby-headphone-like (spatialization) effect to work for all sounds!! (that was the pulseaudio module)
Touchpad gestures like 2-finger scrolling
Screen brightness didn't adjust automatically
[In windows, I could keep the battery charged at 40%-50%, to increase its lifetime. There was no way to set the charge threshold in Linux]
I play some Starcraft II. I'm sure I could make it work on Linux, after 5 hours, with worse visual quality..
The connection to wireless networks was flaky. Something about the applet being started before I had unlocke the keyring.
The system became un-responsive when I transferred large files or did backups.
Did I say the video quality was bad? OK, but it's so bad I have to say it twice! Not just tearing, also dropped frames.
While my laptop cost about as much as a mid-range Mac Pro, I didn't like the idea of lowering the specs just to get an OS I couldn't even try before I bought it. And I was put off Apple by QuickTime on Windows having greyed out menu items that ask you to buy the pro version (no, really! I can't stand software that's "in your face" about selling you stuff)
And this from a Lenovo, which I checked in advance that it should have good Linux support. I'm sure I forgot a few things as well, it has been ~8 months. And it's not actually "keeping me on Windows", I don't really use the laptop all that much, and I use BSD on my desktop and Linux at work. This will probably never be read, but it felt kind of good to get that rant off my chest;)
That's not really 'valuable' to me. But to each their own.
I never thought about it that way, but you're right, I do value spending time with myself and putting together software or hardware, and I do for example find sports to be generally uninteresting The benefit of having something custom made is also greater for people who enjoy using computers a lot (for programming, entertainment, etc), and this is distinct from the mild excitement of building it and the feeling of accomplishment/ownership (the latter I will reluctantly admit to having a little bit of, but I don't think it's very significant).
Nobody would complain about FTP traffic being slowed during busy bursts to avoid interfering with voice traffic.
There are two distinct cases: a) if I pay for 5 Mbit/s and my FTP downloads are being slowed to an average rate of 4.8 Mbit/s I would maybe complain. b) if my FTP speed varies between 3 Mbit/s and 7 Mbit/s over periods of a few seconds because other people are loading web pages, but averages at 5 Mbit/s then that's a well managed network.
(a) can't be fixed by "net neutrality" because that just screws it up for everyone instead of just for the FTP user. The problem is a too great oversubscription factor. So my point is that shaping based on protocol is fine. Another poitn is that the overall share should be the same for each user. I shouldn't be able to run my FTP downloads on a VOIP port and take out the neighbourhood's internet.
PC-BSD is built on FreeBSD, and does a great job of packaging up common desktop functionality. It comes with out of the box support for webcams , flash, sound, and a preconfigured desktop environment. It also has many common applications available as binary packages using its own package manager (including Skype, for example). When you need more applications you can turn to the ports tree, which is somewhat behind the common linux distros, because there is a long UPDATING file with manual tweaks necessary for updating. If you refrain from installing from ports, there are automatic binary updates in PC-BSD. Version 9 also supports automatic in-place upgrades of major versions. On the bad side, the support for sleep and hibernation is entirely reliant on the BIOS, meaning that it doesn't work (I've asked on the forum for a motherboard that supports ACPI S3 sleep, but I got no reply). There is no 3D acceleration with AMD graphics cards, so no dekstop effects are available (I've seen on the IRC that someone may be working on this). NVidia cards are supported using a binary driver, and comes with PC-BSD of course. The installer is very easy for any setup that's less complicated than ZFS mirroring + full disk encryption.
If they had used Amazon or other good cloud hosting service, it would had scaled automatically.
Not if it happened in the middle of the holiday shopping season, when Amazon needs many of their servers for themselves. The hosting is not unlimited at all, just a large pool of servers.
The tools to make a bootable USB-stick aren't quite there (or weren't there 4 months ago at least). On any OS (not win XP) there is a built-in way to burn an ISO. There is no corresponding way to do it for an USB stick. Even if you can use dd on Unix-like OSes, the ISO is still the king of the bootable images. See for example http://www.memtest.org/#downiso. What would you get if you wanted to boot from an USB-stick, and was using a FreeBSD desktop? I didn't try the "pre-compiled binary", since I knew the ISO would work...
At a 5 GB/month cap, your average download speed is 16 kbit/s. Nobody, or at least no geek, should be buying those things. I'm quite sure it doesn't help if software vendors, video sites, etc reduce the size of their downloads to accommodate these caps, reducing the quality for everyone else. The Internet was a great platform for experimentation and a great tool for thousands of purposes, and we geeks got extremely lucky that it also had great commercial uses.This lucky streak seems to be coming to an end, and we may have to start paying more to avoid ridiculous caps and restrictions like blocked ports, IPS-level NAT, etc. On the other hand, video streaming could come to the rescue as a service that "normal" people could demand, at least for the download bandwidth.
As would an embedded public key crypto infrastructure inside the TCP system supporting multiple protocols. And multiple selection of hash checking protocols. Lets make setting a md5 hash at the BGP level obsolete?
No need to do it in TCP when you have IPsec! Unless of course you want per-process authentication instead of per-host authentication -- then you could use TLS. I think you are suggesting a built-in version of TLS anyway, The key management would be a pain if we didn't go with the same error-prone trusted CA model. Windows actually does something interesting here with the "homegroup" system: they use IPSec and IPv6 transparently to get secure LAN communication.
I'm interested in their thoughts for implementing this.. It's easiest to implement in software, but they can't force everyone to use a specific version of Windows or Linux or OS X (I sure hope!!). I doubt they would develop proxy software for FreeBSD, for example. If it's in hardware, then I'm afraid you will get a big slow-down: most computers are connected to the LAN with the same interface as the WAN, so does it have to check and log all the files you transfer to your NAS? If they do destination-based filtering, then people with multiple public IPs are SOL, and anyone can use a local proxy anyway.. Well, unless that also has this tracking HW of course. This just isn't going to happen, because it screws up too much of IT.
I know that's not your point, but I'd recommend going with an AMD processor, since practically all AMDs support ECC memory, and ECC memory is only slightly more expensive than non-ECC.
I figured it out!! (a bit late): In addition to having your main encrypted partition, add a separate unencrypted partition full of tracking software! Then make the TrueCrypt (or what ever) boot loader say something like "Press Esc to boot" and if one hits Esc it loads the unencrypted partition, while if one types the password it loads the main encrypted OS. This gives you the best of both worlds, and I'm seriously considering doing this (installing some old copy of Vista or something for the unencrypted bit). And of wipe the computer clean when getting it back of course.
It's an interesting question whether to use full disk encryption, or to use some kind of tracking and allow the thief to log in so they don't wipe the computer. I suspect that tracking + some kind of lesser encryption (or even just a login password for the "secret" account) is best suited for most people's threat models.
User has to install plug-in to run it. Then it's no better than installing an application.
Web tech has been powerful enough to read the entire frame buffer. That's scary. The idea of "don't install applications" becomes less useful as a security dogma, because websites will be just as powerful as apps. The next iteration of the personal firewalls will have to selectively enable web APIs (like NoScript, but less annoying).
What will be the impact on society when the common kid or inquisitive adult no longer have access to a computer they can program? I can see this going a few different ways, but all of them curve downwards...
From what I could read, ZFS supports "log" devices, essentially write caches, and "cache" devices which are read caches. The read caches get cleared on each reboot. Neither of these devices support TRIM. On a server this does'nt make much difference, but on a workstation it's a big deal (a workstation doesn't have a constant stream of writes, so TRIM on the log device makes a lot of sense). I was very much considering going with FreeBSD for my desktop for this reason, but I settled for Win 7. With a 60 GB SSD I can't really install anything but MS office and a couple of browser, Python and Java. I'm sure that 50 % of that is never accessd though, so a cache would make much more sense! I would consider paying $300 for just the software if it worked well with TrueCrypt. It's the right way to go until we have multi-TB SSDs for cheap. (would be really cool if OCZ offered this for their old drives, but I guess that' s unlikely)
Firefox is great, I love the history search in the URL bar. I can type something in the middle page's title and it sure shows up! I'd be scared if Google could do the same (since Chrome loads suggestions form the Internet). This strategy from FIrefox is kind of a gimmick, I can only hope that it doesn't take down the whole project.
Well it works fine as a marketing lie apparently. I recently had the joy of upgrading my home PC, and there was hundreds of mid-range graphfcs cards to choose from. I got the AMD mainly because they support open source (or so I thought), and I may want to switch to Linux some time in the future if they can finally get the video rendering up to the same level as Windows.. (I have used laptops with AMD and NVidia graphics card on Linux and I haven't noticed much difference actually. The most recent one was an NVidia card, and I could never get rid of the tearing artifacts on the secondary display)
The default is to randomly drop packets, which effectively shares the bandwidth equally between all TCP streams, and UDP is kind of undefined. At home I have set up "tc" on linux to share the bandwidth equally between my machines (+1 allocation for wi-fi leechers). It's a pain to set up, but it makes *much* more sense to divide bandwitdh by IP address than by TCP stream. Why can't ISP set up something like this?
In tc one can use a "Hierarchical Token Bucket (HTB)" where one can set 1) a "rate", which is the guaranteed rate that a class of traffic (e.g. an IP address) gets, and 2) "ceil" which is the max. bandwidth that the class can use if there is additional available bandwidth. Excess bandwidth is shared according to the relative proportion of the rates. The ISP could set the "rate" to be 1/10 of the advertised speed, or what ever factor they use. Protocol-based shaping and prioritation can then be done within each class, affecting only the relative priority of a user's data, but IMHO this is better to leave to the user (though, it's difficult to do ingress shaping for the user).
ISPs have really expensive routers, why can't they do this? Should be simple actually, much easier than tracking connections. There is no reason that 3 Mbps of Skype traffic from customer A should trump 3 Mbps of torrenting from customer B. Just keep the buffering at the shaping device, and the latency will be small.
The current situation is Intel is slaughtering AMD. AMD hasn't had an architecture update in a long, long time and it is hurting them. Clock for clock their current architecture is a bit behind the Core 2 series, which is now two full generations out of date. Their 6 core CPU does not keep up with Intel's 4 core i7-900 series CPU, even on apps that can actually use all 6 cores (which are rare). Then you take the i5/7-2000 series (Sandy Bridge) which are a good bit faster per clock than the old ones and there is just no comparison
Performance per clock is not that important by itself, and AMD are still doing OK on what really matters; performance/$ and performance/Watt .
Sure, there isn't even a real moral dilemma-- people can choose which games they play, and there is practically an infinite supply of them. -- But:It's lame if the rich kids get to "own" the fantasy worlds of games. It's not like they don't have enough shiny toys IRL. Micropayments just create an uncomfortable tie-in between real life and games, removing the "magic" from it. Games are actually a bastion of fairness and equal opportunity in a world that seems less than fair to some people.
Slashdot Mirror
on a forum that 1)allows posts the size of essays and 2) is plagued by goatse, please don't post shortened links ;)
[Amazon's] download client is two years out of date (Ubuntu 9.10) and does not support 64 bit architectures[...]
Apple is evil, but they always make it as easy as possible to buy from them.
So it's easier to use a Windows client in Wine, requiring a 9-step process to install it?
And this from a Lenovo, which I checked in advance that it should have good Linux support. I'm sure I forgot a few things as well, it has been ~8 months. And it's not actually "keeping me on Windows", I don't really use the laptop all that much, and I use BSD on my desktop and Linux at work. This will probably never be read, but it felt kind of good to get that rant off my chest;)
That's not really 'valuable' to me. But to each their own.
I never thought about it that way, but you're right, I do value spending time with myself and putting together software or hardware, and I do for example find sports to be generally uninteresting The benefit of having something custom made is also greater for people who enjoy using computers a lot (for programming, entertainment, etc), and this is distinct from the mild excitement of building it and the feeling of accomplishment/ownership (the latter I will reluctantly admit to having a little bit of, but I don't think it's very significant).
Nobody would complain about FTP traffic being slowed during busy bursts to avoid interfering with voice traffic.
There are two distinct cases: a) if I pay for 5 Mbit/s and my FTP downloads are being slowed to an average rate of 4.8 Mbit/s I would maybe complain. b) if my FTP speed varies between 3 Mbit/s and 7 Mbit/s over periods of a few seconds because other people are loading web pages, but averages at 5 Mbit/s then that's a well managed network.
(a) can't be fixed by "net neutrality" because that just screws it up for everyone instead of just for the FTP user. The problem is a too great oversubscription factor. So my point is that shaping based on protocol is fine. Another poitn is that the overall share should be the same for each user. I shouldn't be able to run my FTP downloads on a VOIP port and take out the neighbourhood's internet.
PC-BSD is built on FreeBSD, and does a great job of packaging up common desktop functionality. It comes with out of the box support for webcams , flash, sound, and a preconfigured desktop environment. It also has many common applications available as binary packages using its own package manager (including Skype, for example). When you need more applications you can turn to the ports tree, which is somewhat behind the common linux distros, because there is a long UPDATING file with manual tweaks necessary for updating. If you refrain from installing from ports, there are automatic binary updates in PC-BSD. Version 9 also supports automatic in-place upgrades of major versions. On the bad side, the support for sleep and hibernation is entirely reliant on the BIOS, meaning that it doesn't work (I've asked on the forum for a motherboard that supports ACPI S3 sleep, but I got no reply). There is no 3D acceleration with AMD graphics cards, so no dekstop effects are available (I've seen on the IRC that someone may be working on this). NVidia cards are supported using a binary driver, and comes with PC-BSD of course. The installer is very easy for any setup that's less complicated than ZFS mirroring + full disk encryption.
If they had used Amazon or other good cloud hosting service, it would had scaled automatically.
Not if it happened in the middle of the holiday shopping season, when Amazon needs many of their servers for themselves. The hosting is not unlimited at all, just a large pool of servers.
The tools to make a bootable USB-stick aren't quite there (or weren't there 4 months ago at least). On any OS (not win XP) there is a built-in way to burn an ISO. There is no corresponding way to do it for an USB stick. Even if you can use dd on Unix-like OSes, the ISO is still the king of the bootable images. See for example http://www.memtest.org/#downiso. What would you get if you wanted to boot from an USB-stick, and was using a FreeBSD desktop? I didn't try the "pre-compiled binary", since I knew the ISO would work...
At a 5 GB/month cap, your average download speed is 16 kbit/s. Nobody, or at least no geek, should be buying those things. I'm quite sure it doesn't help if software vendors, video sites, etc reduce the size of their downloads to accommodate these caps, reducing the quality for everyone else. The Internet was a great platform for experimentation and a great tool for thousands of purposes, and we geeks got extremely lucky that it also had great commercial uses.This lucky streak seems to be coming to an end, and we may have to start paying more to avoid ridiculous caps and restrictions like blocked ports, IPS-level NAT, etc. On the other hand, video streaming could come to the rescue as a service that "normal" people could demand, at least for the download bandwidth.
No need to do it in TCP when you have IPsec! Unless of course you want per-process authentication instead of per-host authentication -- then you could use TLS. I think you are suggesting a built-in version of TLS anyway, The key management would be a pain if we didn't go with the same error-prone trusted CA model. Windows actually does something interesting here with the "homegroup" system: they use IPSec and IPv6 transparently to get secure LAN communication.
I'm interested in their thoughts for implementing this.. It's easiest to implement in software, but they can't force everyone to use a specific version of Windows or Linux or OS X (I sure hope!!). I doubt they would develop proxy software for FreeBSD, for example. If it's in hardware, then I'm afraid you will get a big slow-down: most computers are connected to the LAN with the same interface as the WAN, so does it have to check and log all the files you transfer to your NAS? If they do destination-based filtering, then people with multiple public IPs are SOL, and anyone can use a local proxy anyway.. Well, unless that also has this tracking HW of course. This just isn't going to happen, because it screws up too much of IT.
I know that's not your point, but I'd recommend going with an AMD processor, since practically all AMDs support ECC memory, and ECC memory is only slightly more expensive than non-ECC.
I figured it out!! (a bit late): In addition to having your main encrypted partition, add a separate unencrypted partition full of tracking software! Then make the TrueCrypt (or what ever) boot loader say something like "Press Esc to boot" and if one hits Esc it loads the unencrypted partition, while if one types the password it loads the main encrypted OS. This gives you the best of both worlds, and I'm seriously considering doing this (installing some old copy of Vista or something for the unencrypted bit). And of wipe the computer clean when getting it back of course.
It's an interesting question whether to use full disk encryption, or to use some kind of tracking and allow the thief to log in so they don't wipe the computer. I suspect that tracking + some kind of lesser encryption (or even just a login password for the "secret" account) is best suited for most people's threat models.
This is a cool proof of concept though.
Who actually cares about those logos anyway?
What will be the impact on society when the common kid or inquisitive adult no longer have access to a computer they can program? I can see this going a few different ways, but all of them curve downwards...
From what I could read, ZFS supports "log" devices, essentially write caches, and "cache" devices which are read caches. The read caches get cleared on each reboot. Neither of these devices support TRIM. On a server this does'nt make much difference, but on a workstation it's a big deal (a workstation doesn't have a constant stream of writes, so TRIM on the log device makes a lot of sense). I was very much considering going with FreeBSD for my desktop for this reason, but I settled for Win 7. With a 60 GB SSD I can't really install anything but MS office and a couple of browser, Python and Java. I'm sure that 50 % of that is never accessd though, so a cache would make much more sense! I would consider paying $300 for just the software if it worked well with TrueCrypt. It's the right way to go until we have multi-TB SSDs for cheap. (would be really cool if OCZ offered this for their old drives, but I guess that' s unlikely)
Firefox is great, I love the history search in the URL bar. I can type something in the middle page's title and it sure shows up! I'd be scared if Google could do the same (since Chrome loads suggestions form the Internet). This strategy from FIrefox is kind of a gimmick, I can only hope that it doesn't take down the whole project.
It the total number of people using a browser grows, then the number of users can grow while the user share decreases
Well it works fine as a marketing lie apparently. I recently had the joy of upgrading my home PC, and there was hundreds of mid-range graphfcs cards to choose from. I got the AMD mainly because they support open source (or so I thought), and I may want to switch to Linux some time in the future if they can finally get the video rendering up to the same level as Windows.. (I have used laptops with AMD and NVidia graphics card on Linux and I haven't noticed much difference actually. The most recent one was an NVidia card, and I could never get rid of the tearing artifacts on the secondary display)
The default is to randomly drop packets, which effectively shares the bandwidth equally between all TCP streams, and UDP is kind of undefined. At home I have set up "tc" on linux to share the bandwidth equally between my machines (+1 allocation for wi-fi leechers). It's a pain to set up, but it makes *much* more sense to divide bandwitdh by IP address than by TCP stream. Why can't ISP set up something like this?
In tc one can use a "Hierarchical Token Bucket (HTB)" where one can set 1) a "rate", which is the guaranteed rate that a class of traffic (e.g. an IP address) gets, and 2) "ceil" which is the max. bandwidth that the class can use if there is additional available bandwidth. Excess bandwidth is shared according to the relative proportion of the rates. The ISP could set the "rate" to be 1/10 of the advertised speed, or what ever factor they use. Protocol-based shaping and prioritation can then be done within each class, affecting only the relative priority of a user's data, but IMHO this is better to leave to the user (though, it's difficult to do ingress shaping for the user).
ISPs have really expensive routers, why can't they do this? Should be simple actually, much easier than tracking connections. There is no reason that 3 Mbps of Skype traffic from customer A should trump 3 Mbps of torrenting from customer B. Just keep the buffering at the shaping device, and the latency will be small.
The current situation is Intel is slaughtering AMD. AMD hasn't had an architecture update in a long, long time and it is hurting them. Clock for clock their current architecture is a bit behind the Core 2 series, which is now two full generations out of date. Their 6 core CPU does not keep up with Intel's 4 core i7-900 series CPU, even on apps that can actually use all 6 cores (which are rare). Then you take the i5/7-2000 series (Sandy Bridge) which are a good bit faster per clock than the old ones and there is just no comparison
Performance per clock is not that important by itself, and AMD are still doing OK on what really matters; performance/$ and performance/Watt .
That's not the reason, but BSD is great :)
Sure, there isn't even a real moral dilemma-- people can choose which games they play, and there is practically an infinite supply of them. -- But:It's lame if the rich kids get to "own" the fantasy worlds of games. It's not like they don't have enough shiny toys IRL. Micropayments just create an uncomfortable tie-in between real life and games, removing the "magic" from it. Games are actually a bastion of fairness and equal opportunity in a world that seems less than fair to some people.