Am 27.10.2016 um 01:58 schrieb Mattia Rizzolo:
On Wed, Oct 26, 2016 at 10:56:12PM +0200, Hilmar Preuße wrote:
Hi,
Tried to implement this according to [1]. I gave me an FTBFS (log attached).
That's configure.in being buggy. I quickly asked to a friend of mine
(jrtc27) whom speaks autotools (I don't really), and in about 5 minutes
he came up with https://paste.debian.net/890189/
Apparently this is a requirement since 2001:
http://git.savannah.gnu.org/gitweb/?p=autoconf.git;a=commitdiff;h=e3e7e0a0afcbb870bec9a34d22fe0750393c40a1;hp=aa8dfa178d82db0fae8f5b0692f74e33763a263d
Unless I'm missing something, this also means upstream did not run
autoheader in the last 15+ years?!
Sorry to bother you again. The generated configure script seems to be
broken or incomplete:
checking for struct statfs.f_type... yes
./configure: line 22589: PR_CHECK_STRUCT_ADDRINFO: command not found
./configure: line 22590: PR_CHECK_STRUCT_SS: command not found
./configure: line 22591: PR_CHECK_SS_FAMILY: command not found
./configure: line 22592: PR_CHECK_SS_LEN: command not found
checking sys/acl.h usability... yes
checking sys/acl.h presence... yes
<snip>
checking whether linking with OpenSSL has complete ECC support... yes
checking for Postgres's PQescapeStringConn... yes
./configure: line 24427: syntax error near unexpected token `Wfloat-equal'
./configure: line 24427: ` PR_CHECK_CC_OPT(Wfloat-equal)'
debian/rules:77: recipe for target 'configure-stamp' failed
make: *** [configure-stamp] Error 2
dpkg-buildpackage: error: debian/rules build gave error exit status 2
debuild: fatal error at line 1376:
The code in question is
if test x"$GCC" = xyes; then
PR_CHECK_CC_OPT(Wfloat-equal)
PR_CHECK_CC_OPT(Wformat)
PR_CHECK_CC_OPT(Wformat-security)
PR_CHECK_CC_OPT(Wimplicit-function-declaration)
PR_CHECK_CC_OPT(Wmaybe-uninitialized)
PR_CHECK_CC_OPT(Wpointer-to-int-cast)
PR_CHECK_CC_OPT(Wstack-protector)
PR_CHECK_CC_OPT(Wunreachable-code)
PR_CHECK_CC_OPT(fstack-protector-all)
fi
PR_CHECK_CC_OPT is defined in m4/proftpd.m4. Not sure if it is read, b/c
the functions above PR_CHECK_STRUCT_ADDRINFO... are defined there too.
Should we open a new bug for all this?
I suggest you bring that upstream, not being able to run
autoreconf -f -i
on a autotools project is IMHO a bug that ought to be fixed.
Well, yes. Something to do in the future, not sure when.
Hilmar
--
http://www.hilmar-preusse.de.vu/ #206401 http://counter.li.org
