Re: Self-indexing class?!

[missing@inch.com (Ken Tilton)]

>At 12:10 PM 5/16/95, Ken Tilton wrote:
>>I have this wacky idea of implementing a dataset as a self-indexing class
>>to store tables of data that constitute the "beef" of my app.
>>
>>2> Would anyone know how much this would break Wood?
>
>Wood doesn't persistently store class slots, so your index will
>not be persistent unless you write code to specifically update
>the disk version of the index. Since you need to explicitly store
>the instance anyway, this shouldn't be a problem.
>

Darn. I was hoping that I could just write out...oh, I guess I can just
write out the hash-table per se, then load it at run-time.

Actually, this is starting to feel silly. Not sure it buys me anything,
since the table entry definitions will end up in a macro (or embedded
language!) anyway, in which case the implementation will be sure to put
everything in the right table.

I think I was just tickled by the idea of a self-indexing class. But then I
incur the overhead of having the "key" in the data as well as the key area
of the hash-table.

Is there a known syndrome of enthusiastic Common Lisp newcomers eating
themselves to death in this candy shop, by going crazy with needlessly cute
practices? :)

>? (bar-class-slot (class-prototype (find-class 'bar)))
>BAR
>?

Thanks. We'll keep gong with the self-indexers in case another advantage
rears its head! I may implement the lookup more generically via slot-value
since I'll have a few different such classes.

Cheers,

Ken