I like to use "big-endian" naming molds (love that term!) to define sets of names that when you alphabetize them place related variables next to each other. (i.e. in a completion menu or browser.)

For example, left_foo and right_foo are little-endian, since the least significant word comes first, so they'll be a long distance away from each other in an alphabetized list.

But foo_left and foo_right are big-endian, since foo is more significant than left or right. So they will appear one after the other in an alphabetized list.

Common suffix words are _x _y _z or _min _max, or _left _right _top _bottom, of even singletons like _enabled _loaded _error etc.

But when you combine multiple dimensions together in names, you need to think of which dimensions are more significant, based on how the variables are used, so use foo_x_min foo_x_max, if the positions are important, or foo_min_x foo_min_y, if the ranges are more important.

Sometimes it's hard to decide or ambiguous, so just try to be predictable and the same as all the other code. Think of which variables should appear closest to each other in an alphabetical list.

And avoid middle-endian or random-endian (or sentence-grammar-order-endian) like the plague. A variable name should probably not be a grammatically correct sentence.

Another really annoying linguistic naming smell is "smurfing," where all of class Smurf's instance variables have smurf_ prefixes. Or where all the classes, methods, or instance variables have an "xyz_" prefix where "xyz" is the name of the project or library. Arrgh!!!