|WikiProject Electronics||(Rated Start-class)|
|WikiProject Computing||(Rated Start-class)|
MS Windows implementation
Not sure if it should be mentioned, but on Windows, the implementation is somewhat nonstandard. When all configured network devices are unplugged, Windows also disallows connections to 127.0.0.1. Consequently, standalone machines requiring a loopback network (those with network-licensed software, for example) must have the "MS Loopback Adapter" installed. This disallows addresses in the range 127.0.0.0/8, complaining "IP addresses starting with 127 are not valid because they are reserved for loopback addresses.". (Hey, it made me laugh.) The loopback interface may be hacked in the registry (under HKLM\SYSTEM\CurrentControlSet\Services\Tcpip\Parameters\Interfaces\blah\IPAddress) to convince it to conform to the standard. --Rfsmit (talk) 18:32, 31 December 2008 (UTC)
Silly edit war over network 0
I don't appreciate the undiscussed reversion of my change which clarified the status of networks 0/8 and 127/8. The claim in the edit comment by Kbrose that 0.0.0.0 "is not a loopback network" is technically true of the _network_ 0.0.0.0/8 -- on one view of what a "loopback network" is, which is really not precisely defined in any of the IP standards -- but if it is meant to imply that 0.0.0.0/32 is "not a loopback address" it's false, both empirically and historically.
Empirically: IPv4 hosts are *required* to loop-back packets addressed to 0.0.0.0. Historically: IPv4 loopback was originally defined much like IPv6 loopback is, using the 0/8 "this network" prefix, but differing in that address 0 meant "this host" rather than address 1 being explicitly allocated for loopback purposes. The use of network 127/8 for this purpose was basically accidental, due to the escape of a non-standards-compliant IPv4 stack which used network 127 for this purpose onto the public Arpanet. It has since been -- grudgingly -- enshrined by the standards, but it is accidental. IPv4 originally had _only_ 0.0.0.0/0 as the loopback address, so it certainly should not be removed from this article; however, the discussion of it (and network 127) should be correct, without applying terms like "this network" to network 127, which is wrong.
The extent of the section was disproportionate to the other topics covered here, and largely duplicative of the information in the localhost section. I added any info that was unique to the material I removed here into that article. — Preceding unsigned comment added by Joeldbenson (talk • contribs) 17:11, 25 April 2013 (UTC)