On Sat, May 11, 2013 at 05:31:22PM +0200, Jérémie Courrèges-Anglas wrote: > Stuart Henderson <st...@openbsd.org> writes: > > > s/by/be in DESCR, build fails for me: > > [...] > > > Making all in test > > gmake[1]: Entering directory > > `/usr/obj/ports/mimetic-0.9.7/mimetic-0.9.7/test' > > c++ -DHAVE_CONFIG_H -I. -I../mimetic -I.. -O2 -pipe -DNDEBUG > > -DHAVE_MIMETIC_CONFIG -MT cutee.o -MD -MP -MF .deps/cutee.Tpo -c -o cutee.o > > cutee.cxx > > mv -f .deps/cutee.Tpo .deps/cutee.Po > > /usr/bin/libtool --tag=CXX --mode=link c++ -O2 -pipe -DNDEBUG > > -DHAVE_MIMETIC_CONFIG -o cutee cutee.o > > libtool: link: c++ -o cutee -O2 -pipe -DNDEBUG -DHAVE_MIMETIC_CONFIG cutee.o > > ./cutee -k -o autocutee.mk ./t.version.h ./t.tokenizer.h ./t.base64.h > > ./t.qp.h > > ./t.circular_buffer.h ./t.file.h ./t.mimemessage.h ./t.composite.h > > ./t.tokenizer.cxx ./t.base64.cxx ./t.qp.cxx ./t.circular_buffer.cxx > > ./t.composite.cxx ./t.directory.h ./t.fieldparam.h ./t.fieldparam.cxx > > ./t.strutils.h ./t.rfc822.h ./t.rfc822.cxx ./t.tree.h ./t.parser.h > > ./t.contenttype.h ./t.contenttype.cxx ./t.utils.h ./t.codec.h ./t.codec.cxx > > ./t.find_bm.h ./t.find_bm.cxx ./t.contentdisposition.h > > ./t.contentdisposition.cxx > > cd .. && /bin/sh /usr/obj/ports/mimetic-0.9.7/mimetic-0.9.7/missing --run > > automake-1.11 --gnu test/Makefile > > Provide an AUTOCONF_VERSION environment variable, please > > automake-1.11: autoconf failed with exit status: 127 > > gmake[1]: *** [Makefile.in] Error 1 > > gmake[1]: Leaving directory > > `/usr/obj/ports/mimetic-0.9.7/mimetic-0.9.7/test' > > gmake: *** [all-recursive] Error 1 > > *** Error 2 in . (/usr/ports/infrastructure/mk/bsd.port.mk:2670 > > '/usr/obj/ports/mimetic-0.9.7/.build_done') > > *** Error 1 in /usr/ports/mystuff/mail/mimetic > > (/usr/ports/infrastructure/mk/bsd.port.mk:2389 'all') > > --- Sat May 11 10:43:56 BST 2013 > > This one is weird, cutee[1] is a tool by the same author, that outputs > definitions suitable for inclusion in a Makefile.am. But each time you > run cutee Makefile.in has to be regenerated; this is wrong imho, > upstream should ship pre-generated autocutee.mk / Makefile.in, and only > update this file if needed (if ! cmp -s autocutee.mk.tmp autocutee.mk; > then ...). Also, I don't think tests should be built and ran in a normal > build, and I don't get why they should be ran at install time either. > > Maybe you could just unplug test/ from SUBDIRS (tests pass on i386) and > discuss this with upstream? > > [1] http://www.codesink.org/cutee_unit_testing.html > -- > Jérémie Courrèges-Anglas > PGP Key fingerprint: 61DB D9A0 00A4 67CF 2A90 8961 6191 8FBF 06A1 1494 >
I could unhook the tests. Also if you don't have automake installed the tests still run fine so a valid Makefile is shipped. It's just part of the process to regenerate them if you already have automake installed, so I could remove that part too. I just sent a diff that just adds autoconf/automake as a BUILD_DEPEND which also fixes the problem since it doesn't really hurt anything to regenerate the Makefile even tho it's not really needed. -- James Turner