tuple.index()

[greg]

Nick Maclaren wrote:

> Nope.  Sorry.  Consider the old model where an I/O list is an ordered
> sequence of strings and agents (effectively procedure calls), with no
> constraints on how those are ordered.  With your specification, that
> is neither heterogenous nor homogenous :-)

I don't see any difficulty. This is a list (homogeneous) of
two-element tuples (heterogeneous).

If you're thinking of representing this as just a flat
list, [str1, agent1, str2, agent2, ...], that's a borked
way of doing it and you deserve whatever philosophical
knots you get into as a result.

> It's a complete delusion, because even the claimed assumption of list
> homogeneity is tantmount to saying that Python doesn't encourage (or,
> arguably, support) ANY way of using mutable heterogenous sequences
> (such as the example above).

It's unfortunate that Python doesn't make the homogeneous/
heterogeneous distinction orthogonal to the mutable/immutable
one. Ideally, yes, there would be mutable and immutable
tuples, and mutable and immutable lists. Perhaps also even
fixed and variable sized versions of these as well.

But that would be eight different data types, which is a
lot of trouble to go to just to give you something that's
precisely tailored to exactly what you want. Some might
say it was overkill.

Python takes a more pragmatic approach, and provides
just two: the tuple, optimised for the case of heterogeneous
*and* fixed size *and* immutable -- and the list, for
everything else.

So it's not really that lists are intended *only* for
homogeneous collections, but for anything that can't be
represented as a tuple for whichever reason.

--
Greg