🚀 go-pugleaf

I think that the reason why REDUCE takes a long time to read in large
polynomials is that it puts them into canonical form as it reads them,
which involves sorting.  Whilst in principle one could write an input
routine that just built the appropriate data structures without putting
them into canonical form, you would still need to canonicalize them before
you could reliably perform any non-trivial operations on them.  Therefore,
I'm not convinced that you would gain much.  For example, I would expect
that Maple would input such polynomials much faster because it does not
simplify by default (I think), but you would still need to canonicalize
(simplify) them at some later stage.

If your input polynomials are expressed as products or powers of simpler
polynomials, then you could speed up the input by setting

off exp;

But then the internal representation will be non-canonical (although still
normal), so beware!

Francis Wright