Re: [lojban] More on lojban programatic semantics: Strong typing and inferencing of types

[Brook Conner <nellardo@concentric.net>]

Robin Lee Powell wrote:
> rlpowell@calum> ll saclu
>  saclu              decimal
>                     x1 (me'o) is the [decimal/binary] equivalent of
>                     fractional x2 (me'o) in base x3 (quantity)
> 
>                     [conversion from fractions to decimal-point based
>                     notation] (cf. namcu, frinu)

Recalling that selbri of all sorts are asserting a relationship, I would
expect that "conversion" here is an example and aid to English readers -
saclu does not necessarily imply action or changing of data.

> rlpowell@calum> ll me\'o
> 
>  me'o      LI       the mex
>                     the mathematical expression (unevaluated); convert
>                     unevaluated mathematical expression to sumti
> 
> So the x1 of saclu is a mathematical expression that we're converting from a
> fraction.  That seems to have nothing to do with variable declaration
> to me.

We're not necessarily "converting" it in a variable-assignment or
type-casting sense - we're asserting that x1 and x2 are equivalent, if
we give both x1 and x2. "me'o" (as I read the above) lets you use a
mekso wherever you could use a sumti. If we replace one or the other of
x1/x2 with "ko", wouldn't that be converting it? As in:

la stokuot. saclu ko  --  Provide the fractional equivalent of the
decimal "la stokuot."
 
> Why not use the mekso variable declaration constructs?

Because not all variables are mekso. A file isn't a mekso. Neither is a string.

Or are you suggesting that we use "operator overloading" kinds of things
for variables that aren't numbers or numerical expressions?

Brook