xkcd today- lovely.
in [UK Mac]
|
Prev: Unsupported Firefox...
Next: It's here (at last)
From: Richard Tobin on 23 May 2010 12:35 In article <85ssglFc7oU1(a)mid.individual.net>, Ian Piper <ianpiper(a)mac.com> wrote: >I find the same with the W3C documents: they confuse comprehensiveness >with comprehensibility. They *may* achieve one of these but definitely >not the other! You should try editing one. You write what seems a perfectly clear sentence, and then someone points out an imprecision or ambiguity which has to be corrected. In a textbook you can make an imprecise statement and clarify it later; if you do that in a standard you get complaints. You have to be explicit about everything. And you also get people complaining about statements which are correct but aren't expressed in the way they'd like. If you, as editor, disagree - perhaps you think their change makes it less clear - then the standards process (which aims for consensus) makes it very time-consuming to proceed in the face of a determined objector. It's often easier to give in over such points, but it doesn't make for readable standards. --- Richard
From: Richard Tobin on 23 May 2010 13:10 In article <1jiyf6e.4li2jfrybamlN%real-address-in-sig(a)flur.bltigibbet.invalid>, Rowland McDonnell <real-address-in-sig(a)flur.bltigibbet.invalid> wrote: >> > I can do IP over LocalTalk here, I think... It's understanding what >> > defines `their own network' that I'm short of at the mo. >> Again, simplifying things: for anyone with a single connection to the >> Internet via a broadband router, "your own network" is every device >> connected via either Ethernet or wirelessly on your side of the router >> (plus the router itself, at least the "inside" half of it). >But what defines that as `your own network'? Note the "simplifying things" in the paragraph above. You could reasonably say that the computers run by Edinburgh University are the university's "own network", but they don't use the reserved addresses. In that sense, they are all part of the public internet. A typical home user on the other hand has just one non-reserved address allocated by their ISP, and that address is used by the router. When they connect to (say) Google from one of their computers, the router replaces the real, reserved address in the packet with the single public address, and replaces the port number with one that it generates, and when it gets a packet back to that generated port number it again fiddles with the packet to put back the original reserved address and original port number. This is "NAT". So in this context "your own network" is the set of machines whose addresses get munged by NAT. >> For example, if your router uses an address of 192.168.0.1 and has a >> subnet mask of 255.255.255.0, >Where does this subnet mask come from? You tell the router when you set it up, though it may well be the default. Because you have 16 bits of address available - 192.168.x.y - you could use a subnet mask of 255.255.0.0 and have the whole 2^16 addresses available to you. Historically, the 192.168.x.y addresses were intended to be used as up to 256 subnets each of up to 256 machines (well, 255 or 254, see below), and your router may annoyingly enforce this. If your router is as annoying as mine (Linksys WAG320N) it won't let you use a subnet bigger than 256 addresses even if you're using one of the reserved address ranges intended for that. >What's wrong with 192.168.0.255 as an address? The last address in a subnet is used for broadcast packets. If your subnet mask was 255.255.0.0 it would be 192,168.255.255. The first address is also not usually used, though I don't think that is absolutely necessary. >> Packets addressed to the same network (using the subnet mask to compare >> the computer's own address to the destination address) > >What's using the subnet mask to do this? The computer's IP implementation uses it when your program tries to send a packet to some address. If the computer's address is 192.168.0.77, and the subnet mask is 255.255.255.0, then any address 192.168.0.x is "on the same network". That is, it ANDs the destination address with the subnet mask, and does the same with its own address, and if they're the same then they're on the same network. >And where does it get the subnet mask from? Typically it gets it from the router by DHCP, though you might set it manually. >And how does anything find out where to send anything >for that matter? That is, how does the software work out that IP >address 192.168.0.X is `that node on the Ethernet which is my computer' >or `that node on the Ethernet which is my laser printer' or whatever. Ethernet cards have a unique 48-bit address built in to the hardware known as the MAC address (nothing to do with Macs). The Address Resolution Protocol is used to find the MAC address corresponding to the IP address of a computer on the local network. That MAC address is the thing that makes it be accepted by the right computer. If you type "arp -a" in a terminal window you will see a list of known IP to MAC mappings. >Yeah, but how do they get there? That is, how does a computer know >where to send these packets? Again, usually it gets it by DHCP from the router at startup. The computer broadcasts a request for an address, and the router sends it a message telling it the correct IP address, subnet mask, default route (ie the address to send non-local packets to, which is the address of the router), and DNS server addresses. And again, it is sometimes configured manually. >I get the idea that in the wider internet, >there are domain name servers which contain a full map (no idea how this >works) so you can always look up that sort of thing. No, that provides the mapping from names (e.g. google.com) to numeric IP addresses. The routing - where to send a packet to get to an IP address - is quite separate. There are several routing protocols used, but they are not usually relevant to the end user, which just uses the default route mentioned above. I'm not going to try to describe how routing works on the public internet because I don't understand it properly. -- Richard
From: Ben Shimmin on 23 May 2010 17:48 Ian Piper <ianpiper(a)mac.com>: [...] > I find the same with the W3C documents: they confuse comprehensiveness > with comprehensibility. They *may* achieve one of these but definitely > not the other! The HTML5 spec looks to be the worst yet. There's no way I'm going to read that all the way through -- I'm just going to buy a book which summarises what's new and interesting. b. -- <bas(a)bas.me.uk> <URL:http://bas.me.uk/> `Zombies are defined by behavior and can be "explained" by many handy shortcuts: the supernatural, radiation, a virus, space visitors, secret weapons, a Harvard education and so on.' -- Roger Ebert
From: Peter Ceresole on 23 May 2010 18:15 Ben Shimmin <bas(a)llamaselector.com> wrote: > > I find the same with the W3C documents: they confuse comprehensiveness > > with comprehensibility. They *may* achieve one of these but definitely > > not the other! > > The HTML5 spec looks to be the worst yet. There's no way I'm going to > read that all the way through -- I'm just going to buy a book which > summarises what's new and interesting. Their problem, which is a very real one, is that those specs are just that; specs. So they have to be full and unambiguous. Readbility must necessarily be a very secondary consideration. Manual writing is a specialised and highly skilled job. They need to get a manual writer onto it. But of course that costs money... -- Peter
From: D.M. Procida on 24 May 2010 02:37
Ben Shimmin <bas(a)llamaselector.com> wrote: > Ian Piper <ianpiper(a)mac.com>: > > [...] > > > I find the same with the W3C documents: they confuse comprehensiveness > > with comprehensibility. They *may* achieve one of these but definitely > > not the other! > > The HTML5 spec looks to be the worst yet. There's no way I'm going to > read that all the way through -- I'm just going to buy a book which > summarises what's new and interesting. Well, what do you need - specifications, or information about the new and interesting? You won't generally find one in the other. Specifications documents need to be comprehensive, rather than readily comprehensible. Daniele |