> I think it would be nice to talk about what oo-ifying does and what
> it does not do. It will avoid comments like the above.
...
> It would be better to say that caution should be exercised against
> assuming any one approach is a panacea for all design goals.
These are valid points. To borrow a phrase from Larry Wall, oop should
"make easy things easy and hard things possible". Hence the idea should
not be to make everything "object-oriented", whatever that means to you.
Operations like arithmetic should be easy, e.g., a=b+c+d+e; where
the arguments are scalars, vectors, whatever. However, inverting a
"10,000,000X10,000,000" matrix is not something one will do every day
(hopefully!) so that functional programming ( big_invert(a,b), say) is not
an undue burden on the programmer to look up and figure out how to use.
On the other hand, forcing a programmer to write something like
assign(a,add(a1,add(a2,add(a3,b,c),d),e));
is grotesque when there exist class libraries to efficiently evaluate
a=b+c+d+e;
Any problems associated with consistent indexing among the arguments can
usually be attributed to *mathematical* sloppyness of the programmer
rather than inherent weakness of the linear algebra libraries.
Robert Pierce
This archive was generated by hypermail 2b29 : Wed Feb 20 2002 - 03:20:05 EST