krtisfranks wrote:
> This definition says that it "attaches "te'ai ni'u pa" to all
> subsequent unit-selbri in the tanru" (or is essentially equivalent to
> doing so). We should reinforce that this means that we do not have the
> "per second per second" issue which arises in English. For example, in
> English, "meters per second per second" means m*(s^(-2)) = m/(s^(-2)).
> (This is even though it could conceivably be construed to be equivalent
to
> just m). In Lojban, the former meaning would simply be "mitre fei'u
> snidu te'ai re" or "mitre fei'u snidu (bo?) pi'ai snidu"; the
> latter, on the other hand, would be given by "mitre fei'u snidu fei'u
> snidu" (see below).
>
> Nesting should probably also be explicitly handled. (The obvious meaning
> is that it reciprocates, which is an involution). In other words, count
> how many "fei'u"s precede the given unit in the relevant construct; if
the
> number is even, then the unit is in the numerator; if the number is odd,
> then that unit is in the denominator.
>
>
> The scope should be elaborated too. I personally think that it should
> group everything previous to it as a unit until it hits (from the right
to
> left) the first number coefficient (units-mantissa), mekso operator, or
> something like "li", "lo", or "cu"; so, "ki'otre tei'a re fei'u"
> means "square kilometers per […]" rather than "km^(2/[…])", which is
> pragmatically absurd and a utility which can be gained by the usage of
> other words; meanwhile, it should continue until the end of the
> units-string - in particular, "pi'ai" and "te'ai" have higher priority
> than this word; so, "mitre fei'u snidu bo pi'ai snidu" is the same as
> "mitre fei'u snidu pi'ai snidu", and both would mean 'm/(s^2)'; the
issue
> is that it would need a terminator (although any mekso operator without
> "bo" would work just as well).


Oh, these words are in selma'o KE. That actually resolves a lot of these
issues.