From: Paul Mensonides (pmenso57_at_[hidden])
Date: 2002-08-11 19:05:50
----- Original Message -----
From: "David Abrahams" <dave_at_[hidden]>
> I don't know why you keep ignoring these two points:
> 1. You get the ability to operate on subsequences with the same algorithms
> 2. You get the ability to operate on typelist-style structures from
> different libraries (e.g. Loki typelists, the types of a boost::tuple.
> Even if you don't buy the need for different sequences, the plain fact is
> that different sequences exist. It can be useful to be able to interoperate
> with them.
I am not ignoring them. The first one is not relevant because you could do that
anyway. The second one has some validity, but I'm thinking of the long term
best way, not a way to hack together a bunch things implemented in nearly the
exact same way with different use of capital letters and a few other details.
You break code now in the infancy of template metaprogramming so that you don't
carry the baggage with you later. I disagree that we need to maintain a great
deal of disparate "legacy" template code and that that should be a major factor
in the design or a brand new library. Presumably one that is fit for
Boost list run by bdawes at acm.org, gregod at cs.rpi.edu, cpdaniel at pacbell.net, john at johnmaddock.co.uk