Re: Re: Mutable Bytes

>>>>> "Mark" == Mark Hahn <[email protected]> writes:
Mark> This will have the disadvantage that Prothon keys must be
Mark> totally immutable.  Do you think this will be a problem?

I agree with Greg that only the key should be required to be
immutable. Conceptually, the reason why we want immutable dictionary
keys is, after all, that we can input some known label, e.g. a string,
number or whatever, and retrieve the information that we stored using
that label. It would be very much counter-intuitive if the label were
to change behind our backs. This is because we tend to think of
strings and numbers by their values, not as objects.

However, for more complex keys, I think that the identity of the
object is more important, i.e. a house object will still be the same
house to me if we move around the furniture a bit or give the porch a
new coat of paint.

Perhaps a bit too simply put, if an object contains more complex data
than I can remember and type in fairly easily, I'll want that object's
ID (or similar) to be used as a hash key, not its data.

I'm all for letting strings be 'mutable until hashed' (put that in
your bon and smoke it! :-)), and setImmutable() sounds like a good
idea. However, as far as dictionary keys are concerned, I think that
immutability is required for an object's dictionary key, the function
that returns it should call setImmutable() itself. Does this make
sense?

Martin

-- 
Homepage:       http://www.cs.auc.dk/~factotum/
GPG public key: http://www.cs.auc.dk/~factotum/gpgkey.txt

-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.2.4 (GNU/Linux)

iEYEABECAAYFAkEGIqgACgkQYu1fMmOQldW0mQCfeguG/JYT1Nf6rEUNoOcPQdWX
tt8AnRFdG/3gsRdIR4JjJ6LsUtYn3KM2
=V5Ys
-----END PGP SIGNATURE-----