[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

TECH: Desperately seeking properties

zoi glibau.

Another longish technical rant.
This one's in two parts, one on properties,
and another on {sisku}.
The latter may result in some modification or clarification
of the place structure, in which case I'll pass the outcome
on to lojbab.


.i loi ka co'e zo'u

mi ji'isku zoi glibau.

Some thoughts on properties.

Round about March we decided that a property
was a predicate function, a function returning a truth value,
so that e.g.

        lo ka da blanu
        lambda x:  blue(x)

As it stands, this only handles boolean properties -
either it's blue or it's not.
Some of the things I think about as properties are
quantitative rather than boolean.
The obvious first thought is to use {ni} for these.

        lo ni tuple da

There are two problems-to-me with this as it stands,
the vagueness of {ni}, and the distinction between
function and predicate.

To take the latter point first, in the boolean case we have
a distinction between

        le du'u da blanu
        The predication some-x is-blue
        le ka da blanu
        the property (of-x):  x is-blue

but we don't appear to have a corresponding way of distinguishing
quantities if we use {ni} for both.

The second problem is that {ni} is quite vague about what
quantity it means, which is fine in simple cases,
and useful in certain other cases (e.g. {leni mi nelci ko'a}),
but there are times when it would be useful to be more specific.

By analogy with {du'u}, I propose that we allow {kau} to be
used to specify the particular quantity of interest.
(It may also have other related uses.)
In fact we can already do this with {du'u}.

        ledu'u tu'okau da tuple mi
        The number of legs I have.

So we could simply use a parallel construction with {ka}.

        lo'e kanba cu zmadu mi leka tu'okau da tuple de
        Goats have more legs than me.

where we might want to clarify the identity of the "bound variable"
with a {de zo'u} prenex.
Another example, where the ability to clarify is more important.

        la kolin. zmadu mi leka da cmima tu'okau te mrilu liste
        Colin is on more mailing lists than me.

I suppose this could be {leni te mrilu liste lo se cmima be da},
but it's a bit strained.

If we allowed {ka ... kau ...} to denote specified quantitative
properties in this way,
it would leave {ni} free for the less-well-defined quantities.

I think I prefer this to the other alternative I can see, which is
to use {ni} irrespective of whether we mean a concrete quantity

        leni [tu'okau da] tuple mi

or a functional one

        leni [de zo'u] [tu'okau da] tuple de

and rely on a combination of context and the inclusion of
some of the optional bits to help the audience understand
what we mean.
And of course these possibilities don't have to be mutually exclusive.


.i pe'udo'u pinka .e'o fi ko

.ni'oke'unai lo ckini je se casnu zo'u
mi ji'isku zoi glibau.

On a related topic.

Another even more recent discussion resulted in {sisku}
being assigned a place structure something like

        x1 searches for (an object possessing) property x2 among set x3

(I note in passing that "property" was already an option for the
x2 of {sisku} prior to that discussion.)

I'm not sure if this was intended to be the only valid way of
using {sisku}.  If so, I can foresee some problems.

One of the things I want to be able to say is

        mi sisku la djan.
        I'm looking for John.

I can just about do this with the above place structure

        mi sisku leka [da] du la djan.

but it's a bit clumsy.

I'm sorry I didn't manage to find time to contribute earlier.
I can't help feeling that {sisku} is a Zipfean contraction of
something like {zukte filenu facki}, and {facki} is in turn
some sort of {co'a djuno}.  Since the x2 of both {facki} and
{djuno} is a {du'u}, this leads me to consider the possibility
of {sisku ledu'u co'e}, where the predication could
include something like {vi zo'ekau} to indicate that we're
looking for (we want to know) the whereabouts of something.

I can think of a number of cases.

        mi sisku la djan.
        I'm looking for John.

        mi sisku le nolraitru
        I'm looking for the King.

In both these cases I'm looking for someone/something specific,
and I expect you to know who/what I mean from the information given.

        mi sisku le catra be la kak. rabn.
        (to .iva'i.e'u mi sisku le du'u zo'ekau catra la kak. rabn. toi)
        I'm looking for the killer of Cock Robin.
        I'm looking for who killed Cock Robin.

Here I have someone specific in mind, and I expect you to know what I mean,
even though neither of us can identify or name that particular individual.

        mi sisku da noi nolraixli
        I'm looking for a princess.

I have a particular person in mind,
although I don't necessarily expect you to know who I mean,
at least not until I've given you more information.

(I originally had
        mi sisku lo nolraixli
here, but I eventually decided that the {ka nolraixli}
was incidental in this situation.  I'm still trying to get
a grip on what the latter might really mean, but it must
be something like "I'm looking for _some_ princesses".)

        mi sisku lo ka nolraixli
        (to .iva'i.e'u mi sisku lo za'i mi djuno fi lo nolraixli toi)
        I'm looking for a(ny) princess.

I don't have anyone in particular in mind.
(An event used to be one of the options for the x2 of {sisku}.
However, in the alternative {za'i} version here, {sisku} appears
to be almost indistinguishable from {zukte fi}.)

I think that {sisku} needs to be polymorphic in x2
(i.e. needs to allow x2 to be either e.g. object or property)
to cover all these cases conveniently.


.ike'u pe'udo'u pinka .e'o fi ko

mi'e .i,n.