Test jacobisvd_8 fails, line 74

Submitted by Jitse Niesen

Assigned to Benoit Jacob

Link to original bugzilla bug (#44)
Version: 3.0
Platform: x86 - x87

Description

See the URL for the failure message. This was mentioned by Carlos on the mailing list; I have the same failure on my computer (Red Hat Linux, gcc 4.3.3, 32-bit Intel x86). It only happens when SSE is turned off.

If you look at the message, you'll see that the 'actual' and 'expected' result are equal up to the printing precision. In fact, they differ only by something of the order 1e-17. Do we really need exact equality here?