#4609 atan2(inf,inf) -> 0

[Barton Willis]

This is wrong--it should either be an error or an atan2 nounform:

(%i1) atan2(inf,inf);
(%o1)                                  0

[David Scherfgen]

But Maxima also says that inf - inf = 0 and inf + inf = 2*inf.
Making such a change for atan2 would mean opening Pandora's box ...

[Stavros Macrakis]

Agreed that inf-inf => 0 is a much more common case.
Perhaps have some sort of structured comment in the code when we notice cases like this.
e.g. ;;; BAD atan2(inf,inf)

[Barton Willis]

A few facts:

1. Running the testsuite plus the share testsuite, the case atan2(inf,inf) never happens. That's good.

2. Currently Maxima simplifies atan2(x,inf) to zero. As long as x isn't an extended real, that's OK, but the code doesn't first check for the atan2(inf,inf) case, and I say that's a bug.

3. Inserting code that catches a few cases such as atan2(ind,inf) = 0 doesn't trigger any testsuite failures and doesn't fix any testsuite failures.

4. Inserting code that catches atan2(inf,inf) -> either an error or an atan2 nounform does not trigger any testsuite failures and doesn't fix any testsuite failures.

[Barton Willis]

Fixed by Commit [d00e00] . Regression tests added to rtest_atan2. Closing ticket.