Wikipedia:Reference desk/Archives/Computing/2007 May 20

= May 20 =

Fastest Internet Connection
What is the fastest internet connection in the world? 68.193.147.179 00:18, 20 May 2007 (UTC)


 * That question is almost as pointless as "What is the largest number?". The internet is not at one point. It is a network of hosts. All single lines of this network have a capacity. The total capacity available between your computer and another host depends on the capacity of the lines between them. No matter how fast the first line between your computer and your provider is, you are always limited by the capacity of the backbone.


 * That's a little unfair, it is not a completely ridiculous question. If he's asking what the fastest standard link is, I believe that's OC-192. Stop being so rude and start signing your comments --Oskar 03:29, 20 May 2007 (UTC)


 * No, it really is. "Fastest network connection" would be a decent question, since two computers can be connected at insane speeds (I believe we're up to the TB/sec level in labs), so fast that their hard drives can't even write the info they're recieving, but the internet has so many bottlenecks and jumps that your connection really doesn't matter, since there's a 99% chance that whoever you're trying to connect to is on a much slower connection. Sorta like buying a Buggati Veyron, even though you don't have anywhere to use it but public streets. -- Phoeba WrightOBJECTION! 04:23, 20 May 2007 (UTC)
 * Don't BITE. (I'm assuming you were the first unsigned comment above as well, apologies if not). Josh Holloway  10:03, 20 May 2007 (UTC)
 * It wasn't. If it were, I would've signed it when I wrote my second post -- Phoeba WrightOBJECTION!
 * The military networks must be pretty dam fast and they probably ironed out all the bottle necks as well. --Lwarf 10:07, 20 May 2007 (UTC)
 * Again though, these are closed networks, where speed is, in theory, unlimited, unlike the internet, where your speed is based on how fast the systems you're using are. For example, i'm capable of 6-10MBps in theory, but in practice it's nearly impossible to break 750kbps, and i'm insanely happy when I break 1mbps, because some servers don't handle information as quickly as my connection does. -- Phoeba WrightOBJECTION! 12:18, 20 May 2007 (UTC)
 * OC-768 (40Gbps) has been standardized for some time now and is commercially available as a line card for the Cisco CRS-1 system.

The fastest connections for consumers (that is homes) are the ones being provided in Japan by NTT. They are giving 100 Mbps for $30. Verizon also offers such speeds in their FIOS but it costs more that 100 dollars. But if you consider corporations or military, it can be in terabytes and does not make sense mentioning because they may be misleading and not comparable. -Jerry Kim Meanwhile did you mean Internet or internet? -Jerry Kim

Internet Core routers such as the CRS-1 support connections up to OC-768 (40Gbps). Such a connection is intended for to connect two core routers, but it could in theory be used to connect an Internet router to a CRS-1 acting as a customer-premise router in a (huge) data center, or to even to a single host computer if someone can build an interface. A typical host computer's fastest "Internet" connection is "10GigE" (10 gigabit Ethernet) so the aforementioned "customer" CRS-1 would have an OC-768 interface to the internet and several 10GigE LAN interfaces to hosts in the data center. Do not try this at home unless you are Bill Gates: these things are a bit expensive. -Arch dude 23:00, 20 May 2007 (UTC)


 * I've always used this list to find this answer; according to that it's OC-256 13.271 Gbps. Considering we are still discovering the joys of ADSL here in South Africa (albeit at exorbitant prices), I would die a happy death after being hooked up with a connection like that! Sandman30s 12:11, 21 May 2007 (UTC)
 * Actually, looking at the list again, the fastest internet backbone is OC-48/STM-16 at 2.488 Gbps. So this is your answer. Anything faster than that would be used for private WANs or LANs, like Citrix mentioned above. Sandman30s 12:16, 21 May 2007 (UTC)

widescreen or not
1)Wide screen laptop or not? Which one do you prefer? why? What are the things to be considered before deciding whether to go for wide screen or normal resolution? 2)Will there be any problem if we minimize the size of all windows in Windows OS horizontally and use it as a normal laptop? -Jerry Kim


 * Well if you wanna watch movies on it, widescreen is better. The trend in all visual media displays is towards 16:9 (even photos on NYTimes.com are often cropped to this ratio), so in the future I imagine widescreen will be the only option. --Cody.Pope 11:35, 20 May 2007 (UTC)
 * Widescreen can be more stable on your lap...if you want to take the risk of putting something hot on your crotch. Coolotter88 11:53, 20 May 2007 (UTC)
 * Widescreen all the way. (generally) widescreen laptops have better resolution than non widescreen ones.--Ryan 15:33, 20 May 2007 (UTC)


 * Playing devil's advocate here... I went widescreen a few months ago and I noticed immediately that most websites only took up the middle third (or even worse, the left third) of the screen. Also, some have tiny fonts by default because they're specified in pixels rather than points or some other resolution-independent unit (although that's not so bad because I can tell Firefox never to use any fonts under a certain minimum size). If this annoys you, stick to 3x4 displays. 1024x768 is a good resolution. —Keenan Pepper 20:14, 20 May 2007 (UTC)
 * That's hardly the fault of a widescreen display and rather the fault of the web designers.

@Cody: Aren't most laptops 16:10, not 16:9?? -JOE


 * For gaming, 4x3 is undoubtedly the best. Personally unless they decide that we need a couple dozen more characters in the written english language, the wasted space on each side of the keyboard is unreasonable. And a full sized numpad is ludicrous (sorry HP) --⁪frotht 04:30, 21 May 2007 (UTC)
 * what kind of gamer can condemn a full size numpad on a laptop? That's a GREAT thing! Maybe you've never played them, but plenty of asian games i've played use the numpad for movement. Definitely enough to have me love it enough that I don't want to live without it. That said, my opinion on this, go with the better monitor, which is usually the higher MP count one. You might run your 1920x1200 powerhouse at full res for a good detailed fragging session, and then turn it down to 1280x768 for day to day things to save your eyes. -- Phoeba WrightOBJECTION! 11:20, 21 May 2007 (UTC)
 * Yea if your main purpose is to go on websites, get a normal aspect ratio one or one that has very high resolution (like 1600px+ wide) so you can have 2 browser windows side by side. --antilivedT 05:49, 21 May 2007 (UTC)
 * Can't you force a 4:3 ratio on your laptop anyway (and have 2 black bars on the left and the right? I mean yeah, you have 2 black bars, but don't a 15 in. 4:3 and a 15 in widescreen have the same height? So in the end, you'd only be losing the option to go widescreen if you bought a 4:3 and just a little off the top?--GTPoompt 12:47, 21 May 2007 (UTC)
 * Well same, you can force a 4:3 display to display widescreen, for example from 1024&times;768 to say 1024&times;640 (and have black bars top and bottom). Either way you are losing a lot of pixels, and unless there's any good reason for it.... --antilivedT 04:12, 22 May 2007 (UTC)

Solution orientated architecture
Does anyone know what Solution Orientated Architecture is and what its benefits are ? Note that I do not mean Service Oriented Architecture

163.156.240.17 11:33, 20 May 2007 (UTC)


 * I would guess that no one knows what it means, and that in fact it doesn't mean much of anything at all -- it sounds an awful lot like someone's empty marketing buzzword. With that said, the closest you could get to an "accurate" definition would probably be from the marketeer that constructed the term.  (Be aware that the proffered definition is likely to be along the lines of "A Solutions Orientated Architecture (tm) is one that maximizes customer value by effectuating streamlined processing of both custom and off-the-shelf application paradigms."  If you could get an honest answer out of one of the engineers at the company as to what they did to implement this "SOA", they might say "we added some memory and a faster processor, and put a new splash screen on it.")


 * When parsing marketing-speak like this, it's often useful to ask what the product in question is trying to differentiate itself from. If this "Solution Orientated Architecture" is so good and so special, does that mean that all of its competitors are trying to deliver something other than solutions?  And if so, what?  Newspapers?  Boat anchors?  Doorstops? —Steve Summit (talk) 16:47, 20 May 2007 (UTC)

Thanks Steve - that's what I thought too - Unfortuately I've got to do a presentation on it for a job interview !!!!!!!!!! 163.156.240.17 17:50, 20 May 2007 (UTC)


 * Urk. Good luck.  (And do they really call it "Solution Orientated Architecture"?  I doubt I could ever work for such a company...) —Steve Summit (talk) 18:02, 20 May 2007 (UTC)

One possible meaning might be that they custom design each computer for that user's needs. So, instead of a general all-purpose computer, they produce game-playing computers, music-copying computers, video-producing computers, etc. StuRat 20:25, 20 May 2007 (UTC)


 * I don't know what it means, but "solution oriented architecture" gets 1300 Ghits. I think it's fully buzz-phrase compliant. -Arch dude 20:48, 20 May 2007 (UTC)

Dear All Thanks for your comments - at least I don't feel so stupid as clearly it's a made-up phrase !. I'll ring the company in the morning and try to get some guidelines on what they mean. (Steve - they really do call it orient ta ted )


 * If any company ever mentioned such a phrase to me in the process of a job interview, I would look for a different company.


 * I'd have a bit of fun with them first: "You mean you're still just using plain SOA ? I can't imagine working for a company that hasn't yet embraced Fully Actualized Solutions Orientated Architectiture." StuRat 06:59, 21 May 2007 (UTC)


 * Marketroid-speak is definitely Greek to me, but here's a data point: I was in a meeting with a top-level management/sales guy at my small company and he said (about the networking thingy we make), "we haven't decided yet how to market it. Is it a product? Is it a solution?  It's definitely not an appliance".  I just smiled and nodded in the same way he does when I discuss technical issues.  --TotoBaggins 07:19, 21 May 2007 (UTC)

DPI in photoshop?
When importing a raw image into Photoshop, one has the option to change the DPI. The only noticeable effect this seems to have is changing "canvas size/image size" measures. Assuming that you don't down-sample the image at all, and you just bring a large 10-12 megapixel image to a photo place and print it fairly large (10x12) with the right technology, does this DPI measure mean anything practically? Or is it simply superfluous meta-data? --Cody.Pope 11:40, 20 May 2007 (UTC)
 * It's, simply, the number of dots per inch. When you print, the larger the DPI, the more dots per inch they'll squeeze on the paper. Josh Holloway  12:01, 20 May 2007 (UTC)
 * Maybe, but that doesn't make sense to me in the context I'm asking about. The printer at a photo place has a native DPI, if I'm giving them an image with a resolution above their native resolution (for a given size) they have to throw out data.  Alternately, if I give them an image that I've specified as say 200 dpi (in Photoshop), but I want at 10x12 and my camera file has more data than the 200 dpi specification for that print size in pixels, would their printers then throw out data to match the embedded DPI?  That's can't be right.  I'm looking for the interplay between actual pixels and embedded DPI as specified by meta-data.  Is there any?  --Cody.Pope 12:20, 20 May 2007 (UTC)


 * DPI is a tricky measurement. By itself it does not refer to the number of pixels in an image, it only relates to how the image is considered when it is printed. Generally speaking you want to start by thinking about your target DPI — if your printer is going to print at 300 DPI, then if you had an image that was 300 pixels wide by 300 pixels tall, you could only print it at the size of 1 inch by 1 inch or smaller without having the pixels become visible and chunky. The printer will only be able to print up to a certain amount of detail -- if you give it more than that, it won't be reflected in the final print job. If the printer had a maximum DPI of 200, and you wanted to print something cleanly at 10x12 inches, you would thus need an image that was at least (10*200) x (12*200) pixels, or 2000 x 2400. DPI does not affect the number of pixels in an image (unless you resize it according to DPI, as is often done in photoshop) — it is really a guide for how much printed space you can get out of a given amount of pixels (depending on the medium — some printers can go much higher than 300 dpi, while things broadcast on monitors are generally only around 96 dpi) and so give you some indication of how many megapixels you'll need for a given photo, or what settings you need when you scan something, etc. --24.147.86.187 13:53, 20 May 2007 (UTC)


 * Hmm, maybe I wasn't clear enough. I understand what DPI is and how it is measured.  I'm interested more in how printers interpret embedded DPI data when the specified print size goes against this data.  Especially given that some images don't have any DPI meta-data. --Cody.Pope 14:15, 20 May 2007 (UTC)


 * Commercial printers just print it at the dpi that fits for its physical size. If you ordered 6&times;4, no matter what dpi you've set it's still gonna print 6&times;4. --antilivedT 05:47, 21 May 2007 (UTC)


 * That means the meta-DPI data is effectively meaningless? Good, that's what I thought had to be the case. --Cody.Pope 06:58, 21 May 2007 (UTC)


 * Well if you just want individual prints, it doesn't matter. But if you want to use the image inside something else like a newspaper or magazine, setting the dpi in Photoshop will mean that the image will display as correct physical size if the software is smart enough to read it (most are). --antilivedT 09:20, 21 May 2007 (UTC)


 * Erg, well QuarkXPress and most if not all Adobe based layout programs allow you to resize within the program.  So then what happens to the DPI meta-data? --Cody.Pope 11:56, 21 May 2007 (UTC)


 * It doesn't matter, it will be resampled to the dpi of the page itself. --antilivedT 04:09, 22 May 2007 (UTC)

Gif images
When the animated GIF image moves, how do I stop the image have those fuzzy pixels that move when the object is moving? Thanks, 86.146.170.43 15:58, 20 May 2007 (UTC)
 * Can you show us an example? — Matt Eason (Talk &#149; Contribs) 17:31, 20 May 2007 (UTC)
 * Those white patches aren't meant to be on his face. He asked me if there was any way to get rid of them. Ty 86.146.170.43 17:53, 20 May 2007 (UTC)
 * That's a bad optimization problem. It'll be hard to revert that. Your best chance is to open the gif in the same program that saved it, then saving it again without optimization. &mdash; Kieff | Talk 04:05, 21 May 2007 (UTC)

Problem Writing in Assemly under Linux
For a class, I need to write a program in x86 assembly that, among other things, displays the contents of two 10x10 matrices after initializing them. This is what I've written:

.data a:       .space 10*10*4 b:       .space 10*10*4 c:       .space 10*10*4, 0 fmt: .string "%d %d %d %d %d %d %d %d %d %d\n\n" .globl main main: xorl   %edx, %edx      /* row # */ inita: movl   %edx, %eax imull  $40, %eax xorl   %ecx, %ecx      /* col # */ initaRow: movl   %edx, a(%eax, %ecx, 4) incl   %ecx cmpl   $10, %ecx jl     initaRow incl   %edx cmpl   $10, %edx jl     inita xorl   %edx, %edx      /* row # */ initb: movl   %edx, %eax imull  $40, %eax xorl   %ecx, %ecx      /* col # */ initbRow: movl   %ecx, b(%eax, %ecx, 4) incl   %ecx cmpl   $10, %ecx jl     initbRow incl   %edx cmpl   $10, %edx jl     initb xorl   %edx, %edx              /* row # */ dispa: xorl   %ecx, %ecx movl   %edx, %eax imull  $40, %eax               /* scaled row # */ dispaRow: pushl  a(%eax, %ecx, 4) incl   %ecx cmpl   $10, %ecx jl     dispaRow pushl  $fmt                    /* call printf */ call   printf addl   $44, %esp /* this is where I would jump to dispa */ xorl   %eax, %eax ret

This all works fine but when I put in code to jump to dispa, it assembles fine, but encounters a segfault when it tries to jump to dispa. Does anyone know why? Paulmunger 19:31, 20 May 2007 (UTC)


 * Are you using a conditional or unconditional branch to dispa (since you commented out the line in question)? I have MASM on my machine, and I'm going to try to assemble this (it's been 2 years since I took my one class in assembly...which hewed strictly to 8086--it's similar, but a few more bells and whistles have been added). –Pakman044 22:11, 20 May 2007 (UTC)


 * Actually, it happens with both kinds of branch, as long as the branch is taken--even a straight jmp causes the segmentation fault.Paulmunger 04:30, 21 May 2007 (UTC)


 * You should step through it in gdb and see which line and iteration is segfaulting. The problem will probably be   or passing bad arguments to printf, since they're the only lines down there that deal with memory.  --TotoBaggins 07:12, 21 May 2007 (UTC)
 * Is it ever properly executing the printf call? --jpgordon&#8711;&#8710;&#8711;&#8710; 14:49, 23 May 2007 (UTC)

Streaming Video
is it possible to save video that you play on the Internet to your hard disk? If so, how do you do it?___J.delanoy 20:07, 20 May 2007 (UTC)


 * Sure, if it's from a video-sharing site like YouTube, javimoya.com could do the trick. If it's like an MPEG or whatever file, use VLC Media Player. Splintercellguy 20:12, 20 May 2007 (UTC)

If you're using Firefox, you could try the UnPlug Extension. If there's media on the page, this will let you download it. Deltacom1515 04:17, 21 May 2007 (UTC)


 * Yes, UnPlug is pretty amazing at that sort of thing. Linky, in case you're interested. --saxsux 16:41, 22 May 2007 (UTC)

Need help fixing my laptop keys
Ok, here's what happened. One day, I was typing, when there seemed to be something that prevented me from pressing the "a" key. I gently shook my laptop and I heard a click clack sound, suggesting that there was a rock or other piece of debris under the key. My laptop keyboard looks like this:


 * http://www.tabletpcreview.com/assets/456.jpg

This is a photo I found on google image search. It's not exactly the same, but it hs the same logos.

I tried to pull the rock or whatever out, but couldn't. So I had to just pull the key out. I collected he key, along with the pieces which connected it to the keyboard. Right now it's just a metal plating with a plastic round button in the middle, where my a key should be. I've tried to fit the pieces back in, but I have no idea how they go. I'm scared of opening another key to see how it works, for fear of being unable to fix that one too!

Can someone help me?--0rrAvenger 23:37, 20 May 2007 (UTC)
 * It would be really tricky to describe without pictures, the keyboards I have come across have two delicate pieces of white plastic that fit into each other in a scissor type arrangement, they have tiny little pins on either side, one side of the scissors fits into the laptop part and the other into the key, the unfortunate but common thing that happens is that one or more of the tiny delicate thin "pins" on the sides that the whole mechanism relies on will break, and then there is no fixing it, only perhaps if you try to pilfer the part from a key you rarely use, like the windows key or something. Vespine 05:12, 21 May 2007 (UTC)


 * Are you talking about the Keyboard layout, which is almost certainly QWERTY, or the actual keys themselves? Most keys i've seen can simply be slid and popped into place, but these are desktop keyboards. -- Phoeba WrightOBJECTION! 11:15, 21 May 2007 (UTC)


 * Because desktop keyboards can be much thicker than laptop keyboards, they use a distinctly simpler design. Laptops almost always use the "scissors" design that Vespine described. The key caps can be removed and replaced, but you have to be very careful and skillful. And yes, stealing parts from lesser used keys (like that useless "Windows" key) is a common practice.


 * Atlant 12:29, 21 May 2007 (UTC)


 * Ironically enough, I use my Windows key more often on Mac and Linux than I do on windows. I<3SUPER -- Phoeba WrightOBJECTION!


 * I assume you use it as your meta key?


 * Atlant 15:59, 22 May 2007 (UTC)

...I've tried to fit the "scissor" things together, but it looks like they are broken. Can anyone recommend an ad hoc method of fixing, preferably involving tape? Aesthetics is OUT of the question at this point.--0rrAvenger 17:57, 21 May 2007 (UTC)
 * Same thing happened to me, except I deliberately took them all off to clean and some wouldn't go back on (oops!) If you still have the rubber cap thing attached (under the scissors design thing) then use superglue - worked well for me. If you don't, get the tiniest piece of Blu Tack and make it fill the sensor-whole, then superglue both ends (the end with the sensor and the end with the key) on. It'll be a bit mushy but it'll work. Josh Holloway  19:14, 21 May 2007 (UTC)


 * Can you detatch the keyboard tray? It might be easier to just get a replacement keyboard off ebay at this point. -- Phoeba WrightOBJECTION! 11:59, 22 May 2007 (UTC)


 * The keyboard is always detachable, although it may take some tools and/or skill to do it. In many laptops, the keyboard also serves as the access cover to the RAM DIMM slot(s) or the hard disk drive. It usually connects via single flexible printed circuit cable that is routed to a connector; operating the connector may require care. Overal, replacing the keyboard is usually a pretty simple task (as they are quite prown to failure).


 * Atlant 15:59, 22 May 2007 (UTC)