integrate of bessel_i sign error?

On Oct. 28, Richard Fateman wrote:
----------------------------
>A paper that describes an alternative to the numerical integration of 
>bessel functions when multiplied by other functions has just been 
>revised ...
>http://www.cs.berkeley.edu/~fateman/papers/byparts.pdf
>This might be of tangential interest to people concerned with this thread.
----------------------

Thanks for the reference.

I would like to find a way to numerically integrate
cos(log(x)/x)/x over [0, 1],  (ans = 0.323367)
and perhaps your approach would work.

Ted