[Bug-gsl] J0(5) with a 64bit machine: bug or feature ?

[Luciano Ribichini]

Hallo gsl team,

first of all thanks to the gsl team.
It is a very fine piece of (scientific) software.

I found a slightly inconsistent behaviour I would like to report about.

The computation of J0(5) as shown in the gsl-ref manual gives, on my machine,  
a slightly different result:
the last three digits are 642 on my machine with the manual report,
for the last three digits, 920

I use a 64 bit machine with Debian 5.0 and gsl 1.10
The compiler is gcc version 4.3.2

The minimal code that produce the inconsistency is the one 
given at the beginning of the gsl-ref manual to compute J0(5).

Should I worry about such slight difference in J0(5) between my machine and the 
value given 
in th gsl-ref ? 
Which one is the "right" value ?
Thanks again for your nice piece of software.

Luciano