Hi, On Mon, Feb 23, 2009 at 3:43 PM, Robert Vollmert <rvollmert-lists@gmx.net> wrote:

Hello,

I'm half-way through rewriting nod.txt to use the terminology in mkgmap's routing code, and I'm looking for input regarding the naming of arcs and links.

In the current version of nod.txt, I called any of the pointers between nodes "links", and called the "arcs" we're writing now "direct links", while the "links" we're not writing yet are called "indirect links". On the other hand, libgarmin uses arcs and links, and mkgmap's source code has just arcs for the moment. Consistency would be nice.

Does anyone have a suggestion for a new term encompassing both arcs and links? I'm finding myself writing a lot of "arc/link", and distinguishing them as "direct something" vs. "indirect something" would make the text a lot clearer. What do you think of using the word "edge" (as in an edge of the routing graph)?

arc = direct edge link = indirect edge

If we ever want to write links, it would probably make sense to base RouteLink and RouteArc in a common superclass, which would also benefit from an umbrella term.

My understanding goes like this. Looking at the whole thing as connected trees. Tree with roads class 0, another with class 1 and so on. Where the node which connects two different road classes is a member of the higher class BUT it's root of the lower class, so nodes in the lower class have a link to it's root node. ----------------N1.1-------------- | N0.1(L1.1)-N0.2(L1.1)------------N0.3(L1.1) | | ====================== | | N0.4(L1.2)----N0.5(L1.2)-----------N0.6(L1.2) | ------------N1.3------------------------N1.2 Trees: T1 - N1.1, ..... other nodes in class 1 tree - class 1 T2 - N1.2, N1.3 T3 - N0.1, N0.2, N0.3 -> class 0 T4 - N0.4, N0.5, N0.6 -> class 0 Where N0.4<->N0.1 and N0.6<->N0.3 connect T3 and T4, but they have different root nodes, the closer one. But the root nodes for T3 and T4 are N1.1 and N1.2 . they have links to the root node of the tree they belong to. So Arcs link nodes, Links link trees, a better naming is welcome for the pointer to the tree's root node. -- have fun, alex