Wikipedia:Reference desk/Archives/Computing/2012 May 10

= May 10 =

Buying a camera
I went to the store today to look at new cameras, and the one which caught my attention most was the Nikon Coolpix S6300. Is this a good camera? And is £150 a good price for it? I currently have a Samsung S1050; is the Nikon a big step up and worth upgrading? I mainly want it for HD video which my Samsung doesn't do. I also like to take close-up pictures of plants and insects, however I read that the Nikon is bad at so called macrophotography, is this true? Thanks for the info and advice! Buyingacamera (talk) 18:18, 10 May 2012 (UTC)


 * Go to a local retailer and demo some. ¦ Reisio (talk) 18:35, 10 May 2012 (UTC)
 * Right, which is exactly what I did; I went to the store today to look at new cameras (played around with them a bit within the confines of the store which glue them onto the table and start beeping loudly if you pull the security cord too much) and the one which caught my attention most was the Nikon Coolpix S6300. However, I am not an expert on cameras, which is why I am asking for a third opinion. Buyingacamera (talk) 19:30, 10 May 2012 (UTC)


 * If you want thorough research, go to a site like Digital Photo Review or Steve's Digi-Cams. If you want good advice from a human, go to a real camera shop, not a mass-merchandiser that  has its cameras on leashes.  -- LarryMac  | Talk  19:52, 10 May 2012 (UTC)
 * I appreciate your advice, however if it were possible for me to go to a professional camera shop I would have done so. There are none near where I live, and I am unable to travel long distances. Nor am I experienced with cameras, so all the massive amounts of detailed information available on the various websites really mean nothing to me because I do not understand it, which is why I am asking here on a page where people who may understand this stuff can simplify it for me and answer my 3 basic questions. I am not asking for particularity "thorough research", at least I don't think I am, just yes or no answers. If it helps, here my questions in bullet-point form;


 * Is £150 a reasonable price one would expect to pay for a Nikon Coolpix S6300?
 * Is the Nikon Coolpix S6300 significantly better than the Samsung S1050, or is the overall deference negligible?
 * Can the Nikon Coolpix S6300 take very close-up shots of things a few centimeters from the lens?
 * Thank you Buyingacamera (talk) 20:04, 10 May 2012 (UTC)
 * From the official Nikon technical specifications for S6300:
 * It sounds like you're quoting UK prices; here in the US, the S6300 should be much cheaper. If you have settled on buying a Nikon point-and-shoot, I would recommend the more expensive, but better S8200.  It's one of the best point-and-shoots on the market.  By my math, £150 is about $250 (US), which is more than I'd pay for a "middle of the road" point-and-shoot like the S6300 - especially when the S8200 costs about the same.  Here in the US, you can find the S8200 for that $250; and it has a bigger screen, a better lens with sharper focus and a lot more zoom, and a few more features to round things out (better metering, better face-detection, better auto-focus).  In any case, if the price is higher in the UK, it's probably more expensive across all brands and models; but you may still consider shopping around a bit more.
 * I have a tendency to look for flaws in photographic images. In Nikon's low-light sample images, I am impressed by the low noise (they use great sensors); but I see chromatic aberration (a mark of a moderate- to poor- lens).  This is "expected" in a $200 (£150) point-and-shoot camera.  I am very picky, and I can justify spending more on a better lens (and an interchangeable-lense camera); but you might not be so picky.  Compared to the Samsung S1050, the Nikon has much better image quality.  The S1050 is also almost 5-year-old technology (it was released in 2007), and it has been discontinued from new production.  Samsung's latest toys in their point-and-shoot camera price-range have dual-screens (one on the front).  It's a fun gimmick (especially if you're photographing portraits of kids - you can animate a dancing bear or whatever and show it on the forward-facing screen, and the kids laugh and smile when you shoot the photo).  That's going to be the deal-breaker for "image quality" to many consumers - but doesn't affect actual image-quality of the Samsung device on technical merits.  I have not found many great full-sized Samsung S1050 images to compare, so I'll decline from commenting on its image-quality specifically.
 * For S6300: the close focal point is approximately 50 cm, when at minimum zoom; approximately 1 meter when fully zoomed. This is not bad for a point-and-shoot lens (though it is not competitive with specialty macro-lenses, or even, say, the iPhone 4S).   There's also a "macro mode," with a purported 4-inch focus distance; but I'll mildly suggest you accept that number with caution.  Even if accurate, you have a 10x optical zoom - so... unintuitively - you will get a "closer" shot by standing at a farther distance and zooming to maximum-telephoto zoom on the S6300.  If you stand close and zoom out, your image magnification will be less (see some explanation, and magnification for some math).  You should not expect the point-and-shoot to compete with much-more-expensive DSLR cameras equipped with specialty macro lenses.  Those lenses have large apertures, primary (fixed) focal-lengths, and very tight close-focus-points; but one Nikkor macro lens costs more than your entire camera; a good Nikkor macro like the AF 60mm is about $600 (~£400 to £500?  I apologize, I don't know how Nikon sets its market prices for products sold in the UK; it's not as straightforward as a direct dollar-to-pound conversion).
 * Hopefully this will provide some perspective. You should get the camera that is most suitable for your needs.  "Good camera" is very subjective.  Personally, I am a fan of the iPhone 4S; but I also carry my Nikon D90, which is too bulky and heavy for most casual users.  Nimur (talk) 00:13, 11 May 2012 (UTC)
 * Awesome! Thank you so much Buyingacamera (talk) 11:41, 11 May 2012 (UTC)
 * (edit conflict)It seems like a good price considering that amazon.co.uk, which normally has lower prices than retailers, is selling the camera for almost £8 more. A quick glance at some online reviews for the S6300 shows that its minimum focusing distance of 10 cm and its light weight makes it very challenging to get good macro shots.  In other ways it appears to be a good point-and-shoot camera, though if ultra-portable isn't a critical factor and you want passable macro capabilities, it would probably be worth continuing the hunt.  P.S. I think in exchange for the advice you should upload your photographs to Wikipedia! ;)  Julia\talk  20:18, 10 May 2012 (UTC)

Either the size/quality looked good to you or not. The technology is advanced enough that there is a lot of room for preference alone. ¦ Reisio (talk) 22:38, 10 May 2012 (UTC)

Networking - ICMP
Is the ICMP protocol in the same layer as IP or is it encapsulated in IP? --TuringMachine17 (talk) 20:35, 10 May 2012 (UTC)


 * Internet Control Message Protocol says "Although ICMP messages are contained within standard IP datagrams, ICMP messages are usually processed as a special case". -- Finlay McWalterჷTalk 21:24, 10 May 2012 (UTC)
 * The reason I ask is because I have a general understanding of networking and I know that a protocol in one layer should only communicate with thos directly above/below it. Wikipedia and other sources describe ICMP as being in the network layer (as is IP) and seem to be used together as I have seen when using wireshark. Is this because TCP/IP dosn't follow the OSI model strictly. --TuringMachine17 (talk) 21:47, 10 May 2012 (UTC)


 * Indeed, TCP/IP doesn't map very well to the OSI model (and it only gets more complicated when more levels of encapsulation are used). And lots of network equipment sneaks a peek at other layers (e.g. ethernet routers may examine the IP headers of the packets they're forwarding and do "helpful" things) which again breaks the neat OSI layer-by-layer view. -- Finlay McWalterჷTalk 21:57, 10 May 2012 (UTC)


 * ICMP is encapsulated in IP. It's at the same level as TCP and UDP. I have no idea why the Wikipedia article groups it with IP instead of TCP. Probably that should be fixed. -- BenRG (talk) 00:17, 11 May 2012 (UTC)


 * ICMP is really the control plane for IP (it would have been better if it had been called IP-CM instead) rather than a layer above it. IP needs ICMP to function - ICMP manages reachability, flow control, and route advertisement and redirection - all for IP datagrams. Internetworking with TCP/IP by Douglas Comer puts it well "It is important to keep in mind that even though ICMP messages are encapsulated and sent using IP, ICMP is not considered a higher level protocol - it is a required part of IP. The reason for using IP to deliver ICMP messages is that they may need to travel across several physical networks..."  This built-on-but-somehow-irrevocably-wedded-to problem is an artefact of the very layering worldview that OSI espouses.  When folks talk about the same stuff in ATM they wisely use a three-dimensional view like this which acknowledges the reality that for each layer there is an in-layer control plane sending management traffic along side the normal "user" traffic of that layer. -- Finlay McWalterჷTalk 12:37, 11 May 2012 (UTC)


 * Here's what I hope is an apt illustration of ICMP's incestuous relationship with IP; how ICMP echo-request packets are handled by intermediate hosts. If I send an ordinary IP packet to google.co.uk(74.125.132.94), I'm making a one-off end-to-end communication through IP (where I can afford to be ignorant of the route) and I'd expect only to receive a response from 74.125.132.94; intermediate IP routers are expected to only read the IP headers of my packet, and to route it accordingly (or just drop it on the floor if they can't handle it) - in a strictly layered view, they've certainly got no business inspecting the data encapsulated in that packet and changing their behaviour on that basis - and they certainly don't have any business sending me a reply. But if I send 74.125.132.94 an ICMP_ECHO packet with a TTL of say 7, I do get a response from one of the intermediate machines:

> ping -t 7 google.co.uk PING google.co.uk (74.125.132.94) 56(84) bytes of data. From 209.85.253.92 icmp_seq=1 Time to live exceeded
 * Who is this 209.85.253.92 character? I've never send him anything, yet he's sending me an ICMP reply packet. If ICMP was layered strictly on IP, 209.85.253.92 wouldn't have looked inside the IP packet as it passed him (just as he doesn't look inside TCP or UDP packets). But to response it's clear that 209.85.253.92 did look inside the packet at the encapsulated data. In fact, when he received a packet with its TTL of 1 (which means it would be decrementing it to 0 itself) he checked the ip:protocol field, saw that it was 0x01, which he knows specifies ICMP. Then he broke the encapsulation to open the ICMP header and read the icmp:type field. In this case he saw that value was 8, an ECHO REQUEST packet. So he dropped the packet (which a pure IP system would do anyway, as the TTL was expired), but he also generated an ICMP Time Exceeded packet and sent that back to me (other IP packets, and other kinds of ICMP packets, wouldn't). So this shows that ICMP isn't layered strictly on IP, but that it gets special sideways access to the IP layer, which real users of IP like TCP and UDP don't. -- Finlay McWalterჷTalk 17:11, 11 May 2012 (UTC)
 * Except that ICMP Time Exceeded, as the article itself indicates, can also be returned for TCP and UDP packets. So the routers really don't need to looke 'inside' the packet just for this. traceroute (except the Windows version, which uses ICMP if I'm not mistaken) and tcptraceroute are built on this principle. Unilynx (talk) 20:59, 11 May 2012 (UTC)