Re: trouble with bessel function test ...

[Brian Gough <bjg@network-theory.co.uk>]

Achim Gaedke writes:
 > I ran specfunc/test in gdb:
 > 
 > PASS: gsl_sf_bessel_Jnu_e(0.0001, 1.0, &r)
 > 
 > Program received signal SIGSEGV, Segmentation fault.
 > 0xa6058 in gsl_sf_bessel_JY_steed_CF2 (nu=0.0001, x=0.10000000000000001,
 > P=0xffbef428, Q=0xffbef420)
 >     at bessel.c:628
 > 628       double den = br*br + bi*bi;
 > 
 > Maybe, this way of initialisation is not possible in pic code...
 > The gcc version was 2.8.0.

It's valid C ...  beyond that I don't know what's causing it.

 > I've tested it with egcs-1.1.2, there was no error in specfunc, but a
 > compile error at
 > /zpr/gcc/egcs1.1.2/bin/gcc -DHAVE_CONFIG_H -I. -I. -I.. -I.. -I..    -g
 > -O2 -c test.c
 > /usr/ccs/bin/as: "/var/tmp/ccaQ2xka.s", line 2207: error: statement syntax
 > /usr/ccs/bin/as: "/var/tmp/ccaQ2xka.s", line 2343: error: statement syntax
 > 
 > maybe there is a problem with egcs...

I think so ... GSL won't be responsible for problems in .s files, that
is a compiler bug.

 > With linux, I tried to build gsl-0.8 in a seperate build tree, but failed.
 > Has anyone tried too?

It works fine for me.  What error do you get? (and is it compiler
related?).