0.17.4 make check failure: t.grftest

[Matthias Andree]

On Tue, 23 Mar 2004, Matt Garretson wrote:

> David Relson wrote:
> >Matt Garretson wrote:
> >>>  8   S 1.000000   S 1.67e-15
> >>Could it be a rounding error of some sort?  My GSL is gsl-1.1.1-5,
> >An error of 1.67e-15 is pretty small, no?  However it shouldn't be
> >there.  Have you tried with the gsl code that's included with
> 
> 
> Somehow i'd overlooked the "--with-included-gsl" configure option
> till  now.  I just rebuilt with it, and now all the tests pass
> fine.  Sorry about that, and thanks, David & Matthias, for the
> insight.

No need for apologies, it is important that such differences be reported
early so we can fix them early. While numeric mathematics are sometimes
difficult and the deviations are miniscule, a dependency on the exact
GSL version is undesirable (barring bugs in particular versions GSL, of
course, but none of these that might affect bogofilter have surfaced
over the whole time we've been using GSL).

-- 
Matthias Andree

Encrypt your mail: my GnuPG key ID is 0x052E7D95