Re: Strange problems with data-tests.el

[Eli Zaretskii]

> From: Pip Cet <address@hidden>
> Date: Mon, 13 Aug 2018 10:06:12 +0000
> Cc: address@hidden
> 
> On Sun, Aug 12, 2018 at 6:38 PM Eli Zaretskii <address@hidden> wrote:
> >   (< 0.5 most-positive-fixnum (+ 1.0 most-positive-fixnum))
> 
> I think that test is a bit misleading, as (< most-positive-fixnum (+
> -128.0 most-positive-fixnum)) also returns true.

Not on a 32-bit host with a 32-bit EMACS_INT, it doesn't.

> It's not the addition that makes the RHS greater than the LHS, it's
> the rounding that happens when it's converted to a float.

With 64-bit EMACS_INT, yes.