SUMMARY: Re: Compile error V0.91.1: gsl_sf_exp_e redefined

[Matthias Andree]

On Thu, 10 Jun 2004, Uwe Richter wrote:

> - C compiler is:
>   $ cc -V  # Compaq C V6.3-027 on Compaq Tru64 UNIX V5.1 (Rev. 732)
>            # Compiler Driver V6.3-026 (sys) cc Driver
>   The system's native cc produces much more faster code than gcc -
>   so for "primary service binaries" gcc is no option for us.

This compiler should have an option to switch floating point code,
probably -ieee_with_no_inexact, according to GCC documentation.

Does your compiler support -ieee_with_no_inexact?
man cc(1) or other forms of Compaq C or cc documentation should tell you.

As a quick check (that should make the t.u_fpe test pass if it works), try:

rm -f config.cache
env CFLAGS="-O2 -ieee_with_no_inexact" sh configure [other options go here]
make all
cd src/tests
make check TESTS=t.u_fpe

Should give a PASS: t.u_fpe

> - bogofilter version:
>   0.91.1 with patches received from
>   Matthias Andree <matthias.andree at gmx.de>
>   (see attach check-extern_inline.diff)
>   Thanks!

You're welcome, and thanks for testing.

I was wondering whether today is a holiday in Thuringia and I had to
wait til Monday for the result, but it appears not. :)

> - $ make check # (see attach make_checl.log)
>   failed for:
>   - t.u_fpe: ./t.u_fpe[7]: 498301 Floating exception
>              maybe as Alpha is little endian
>   - t.multiple.wordlists: ?
>   - t.MH: ?
>   - t.maildir: ?
>   scipped:
>   - t.valgrind: ?

Some of the tests are known to be broken (they only differ in the
version=0.91.1 tag), patch "update-testsuite.diff" attached.

> >Compiling try.c (cc -o try try.c) should work, and running ./try should
> >print 1764.
> 
> This worked without errors for the system's cc mentioned above.

And it tested a different aspect so it didn't figure what we needed.

Can you please send me your config.log file and the output of
./config.guess off-list?

BTW, valgrind is a memory access checker that only works on Linux on
x86 AFAIK (although ports to other systems and machines are in
progress), it's normal that the test is skipped.

-- 
Matthias Andree

Encrypted mail welcome: my GnuPG key ID is 0x052E7D95