hydra-users
[Top][All Lists]
Advanced

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: Nettle, Gnutls, and Solaris.


From: Mats Erik Andersson
Subject: Re: Nettle, Gnutls, and Solaris.
Date: Thu, 30 Aug 2012 16:32:09 +0200
User-agent: Mutt/1.5.18 (2008-05-17)

Dear Ludovic,

lördag den 25 augusti 2012 klockan 16:33 skrev Ludovic Courtès detta:
> Hi Mats,
> 
> Mats Erik Andersson <address@hidden> skribis:
> 
> > We can all agree that the showstopper for guile-2.0.* lies in
> > libunistring, which on OpenSolaris throws 27 cordumps out of
> > 407 test cases.
> 
> Hmm, not on Hydra (OpenIndiana), AFAICS, where it???s really Guile???s test
> suite that fails.

Yes, I am also running on OpenSolaris 2010.03 in order to get as many
indications as possible. OpenIndiana works better, but fails on four
tests in libunistring-0.9.3. One error is avoided using

  ./configure SHELL=/usr/gnu/bin/sh      (or /usr/bin/bash)

since

  /usr/bin/sh --> /usr/bin/i86/ksh86

on OpenIndiana, as well as OpenSolaris, fails to use the idiom

    ${1+"$@"}

as POSIX and GNU expect when the sole argument $1='' is empty.
This is crucial in one libtool scripted test, whence this one
test case fails with the naive configuration.

> 
> > Let me therefore propose a build task gnutls-without-guile,
> 
> I???ve done that.
> 
> Now GnuTLS fails to find Nettle:
> 
>   checking whether to use nettle... yes
>   checking for libnettle... no
>   configure: error: 
>     ***
>     *** Libnettle 2.4 was not found. You must compile nettle with gmp support.
> 
> (From <http://hydra.nixos.org/build/2967876/nixlog/1/raw>.)

Judging from the printout, you have omitted every command line switch
to "./configure". No wonder you have troubles reaching the set goal!

On my OpenIndiana system, gnutls-3.0.22 works fine with

  ./configure CPPFLAGS="-I/usr/include/gmp -I/usr/include/idn" \
       --with-libnettle-prefix=/export/home/mats/roen
 
Remember LD_LIBRARY_PATH in case of need for all tests. As we have learned
already, libnettle and also Simonsson's work are picky about the location
of <gmp.h> and <idna.h>, so they need manual help in CPPFLAGS.

The tests are passed except some stalled network communication where
gnutls-srv and gnutls-cli are waiting for each other. I have not got
around to debug those.

Best regards,

  Mats E A



reply via email to

[Prev in Thread] Current Thread [Next in Thread]