I'd rather think that semantic overloading is nothing you can escape from. Whatever terms we might define, they will bear more semantic significance over time and over practice. No truth is true forever. So, to me, semantic overloading of 'address' may not have been the real evil we had to get rid of. -- DY Well spoken. So any correction of semantic overloaded definitions is just the attempt to make the own interpretation the prevailing one. Honestly: I am worried that the pretty new term "Locator" is going to be bent, too - just like the term address has been. Locator (just like the address ought to denote) denotes the location of a node. (which means WHERE that node is). And there are better and worse locators. If the locator contains "fleet street" you cannot infer what is the neighboring street. However, given the locator contained "7thStreet" you may conclude that 6th Street and 8th Street are next to it. Heiner -----Ursprüngliche Mitteilung----- Von: Dae Young KIM <[email protected]> An: William Herrin <[email protected]> Cc: IRTF Routing RG <[email protected]> Verschickt: Sa., 12. Jun. 2010, 3:17 Thema: Re: [rrg] Terminology straw poll On Sat, Jun 12, 2010 at 9:18 AM, William Herrin <[email protected]> wrote: >> So, are you saying this semantic overloading of 'address' is the >> reason to look for newer terms? Or, are you saying that such semantic >> overloading is rather natural. > > Hi DY, > > If was making any points, and I'm not entirely sure I was, it would > probably be these: > > 1. Terms like "host," "address," and "router," already have common and > well understood if not well defined meanings. While it might be > valuable to better formalize those meanings this group, both singly > and as a whole, lacks the prerogative to -change- those meanings. > > To the extent that Ran's definitions change those terms, the > definitions are wrong. For example, any machine that speaks IP is a > host. That includes routers. I'm not saying that should be the case, > I'm saying that IS the case as the word is actually used. If we think > it would be valuable to define a type of machine which originates and > accepts IP packets but will never under any circumstances forward one > elsewhere, we should pick a new word. Yes. I'd agree. In put some other words, a node is one involved in IP packets. If they're stub-nodes, they are also usually called host. If they're relay nodes, they are also interchangeably called routers. But, in an exact sense, there's one and only one type of machine, 'node'. Host and routers are, in a way, only loose synonyms used interchangeably with 'node', for conversational convenience' sake. > 2. Redefining well understood and commonly used words in terms of > poorly understood theoretical constructs is an exercise in getting > other engineers to roll their eyes at you. An address is what it is, > and that doesn't include anything about locators and identifiers. On > the other hand, it might be valuable to define a locator (which is NOT > a commonly understood term) as an address used only in a particular > way or (if it's the group's preference) an address whose use includes > particular functionality. > > Quite frankly, I think it would help the discussion to have two such > terms - one for an address that is ONLY used in the next hop selection > process and a different one for an address whose use INCLUDES the next > hop selection process. Otherwise we'll keep interchangeably and > ambiguously using the term locator to mean whichever one we happen to > want to discuss. Maybe. I'd rather think that semantic overloading is nothing you can escape from. Whatever terms we might define, they will bear more semantic significance over time and over practice. No truth is true forever. So, to me, semantic overloading of 'address' may not have been the real evil we had to get rid of. -- DY _______________________________________________ rrg mailing list [email protected] http://www.irtf.org/mailman/listinfo/rrg
_______________________________________________ rrg mailing list [email protected] http://www.irtf.org/mailman/listinfo/rrg
