Re: Supporting 2u

[Ilya Zakharevich <nospam-abuse@ilyaz.org>]

On Mon, Aug 12, 2024 at 12:48:46PM +0200, Vincent Lefevre wrote:
> On 2024-08-12 12:04:36 +0200, Ruud H.G. van Tol wrote:
> > >    P.S.  IMO, 2u should better be parsed (as expected by the user!).
> > 
> > Rather not. For example, 2e1 is already syntax, so can't also mean 2*e1.

This does not stop other ambiguous syntaxes — like ++.

> Agreed. qalc (Qalculate) is quite horrible in this respect:

It is not very clear to me how existence of a piece of software which
implements it horribly may be an argument — for or contra!

Of course e and p MAY give confusing results.  So when run-ins are
supported, there should be a warning configurable to trigger

  never / always / on-floating-point-e\p / on-a-variable-e\p / whenever-both-variants-have-already-been-used.

(Maybe even disable the “never” choice?!)

Hope this helps,
Ilya