Hi Thorsten,
> as far as I understand, property lists in PicoLisp are kind of reversed:
>
> ,----------------------------------------------
> | ((val1 . key1)(val2 . key2) ...(valN . keyN))
> `----------------------------------------------
Correct. The value of a property is in the CAR of the corresponding
cell, to allow it to be treated as a 'var' and thus be passed to
place-modifying functions ('push', 'set', 'inc' etc.).
This can be done with the 'prop' and '::' functions. For example, to
exchange two properties 'x' and 'y' of a symbol 'A':
(with 'A (xchg (:: x) (:: y)))
> Are plists only used as property lists of symbols, or can they exist
Property-"list"s are an *internal* representation of value-key-pairs in
symbols. They don't actually exist in a physically accessible form.
See e.g. the "Symbol" description in "doc64/structures":
Symbol
|
V
+-----+-----+ +-----+-----+
| | | VAL | |'cba'|'fed'|
+--+--+-----+ +-----+-----+
| tail ^
| |
V | name
+-----+-----+ +-----+-----+ +-----+--+--+
| | | ---+---> | KEY | ---+---> | | | | |
+--+--+-----+ +-----+-----+ +--+--+-----+
| |
V V
+-----+-----+ +-----+-----+
| VAL | KEY | | VAL | KEY |
+-----+-----+ +-----+-----+
The property-cells are part of the symbol's internal structure.
We can see from the above that while a symbol's TAIL cell may point to a
property list, this is not a normal NIL-terminated list, because it may
be followed by the symbol's name.
You can genearate a list of value-key-pairs from a symbol's properties
with 'getl', but this will always return a freshly cons'ed list, not
identical to the symbol's internal cell structure. In the above example,
the returned list would be something like ((VAL . KEY) KEY (VAL . KEY))
> Are plists always built with cons cells and alists only sometimes?
Sure. Everything in PicoLisp is built with cells. Why should association
lists be only sometimes so?
♪♫ Alex
--
UNSUBSCRIBE: mailto:picolisp@software-lab.de?subject=Unsubscribe
