1.8.0 changes behavior of == in Comparable

ruby-talk
1

Running my unit tests with 1.8.0, I noticed that the behavior of == in
Comparable has changed from 1.6.8 to 1.8.0. In 1.6.8, if <=> raises an
exception, == catches it and returns false. In 1.8.0, == doesn’t catch the
exception.

This change seems reasonable, but I’d be interested in learning more about
the reasoning behind it. Anybody care to comment?

Thanks!

2

See [ruby-talk:40612] and the following thread (particularly
[ruby-talk:40625] and [ruby-talk:40772]).

Paul

···

On Fri, Dec 27, 2002 at 12:08:21AM +0900, Tim Hunter wrote:

Running my unit tests with 1.8.0, I noticed that the behavior of == in
Comparable has changed from 1.6.8 to 1.8.0. In 1.6.8, if <=> raises an
exception, == catches it and returns false. In 1.8.0, == doesn’t catch the
exception.

This change seems reasonable, but I’d be interested in learning more about
the reasoning behind it. Anybody care to comment?

3

Hi,

···

In message “1.8.0 changes behavior of == in Comparable” on 02/12/27, “Tim Hunter” cyclists@nc.rr.com writes:

Running my unit tests with 1.8.0, I noticed that the behavior of == in
Comparable has changed from 1.6.8 to 1.8.0. In 1.6.8, if <=> raises an
exception, == catches it and returns false. In 1.8.0, == doesn’t catch the
exception.

This change seems reasonable, but I’d be interested in learning more about
the reasoning behind it. Anybody care to comment?

Performance. 1.6 “<=>” method reported its errors by exceptins,
whereas 1.8 “<=>” would report errors by returning nil.

						matz.