[Bug tree-optimization/97909] expr_not_equal_to (mainly in match.pd) vs. ranger

[amacleod at redhat dot com]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=97909

--- Comment #3 from Andrew Macleod <amacleod at redhat dot com> ---
It should be able to access  the currently known global values that have been
computed, or if there isn't one, it could still go and calculate a global
range.   Which is what the default condition would be (ie, if you don't supply
a context).

And we can explore under what conditions we can use more info... I suspect
there are times it would be OK and we can supply a context only then.

Regardless, all the on-demand engine does is give you the same information you
would have had if you calculated all of it before doing whatever you are about
to do.   It just avoids the need to do so for anything that isn't relevant.

And if there are times that we can't calculate ranges, then we make sure we
don't.