Trying to build libmcrypt (2.5.1, no -r), yields me this: configure: running /bin/sh './configure' --prefix=/usr '--prefix=/usr' '--mandir=/usr/share/man' '--infodir=/usr/share/info' '--datadir=/usr/share' '--sysconfdir=/etc' '--localstatedir=/var/lib' '--disable-posix-threads' 'CFLAGS=-mcpu=i686 -O3 -pipe -g' 'LDFLAGS=-g' --enable-ltdl-install=no --cache-file=/dev/null --srcdir=. configure: warning: CFLAGS=-mcpu=i686 -O3 -pipe -g: invalid host type configure: warning: LDFLAGS=-g: invalid host type configure: error: can only configure for one host and one target at a time configure: error: /bin/sh './configure' failed for libltdl !!! ERROR: The ebuild did not complete successfully. !!! Function src_compile, Line -112, Exitcode 1 !!! (no error message) !!! emerge aborting on /usr/portage/dev-libs/libmcrypt/libmcrypt-2.5.1.ebuild . SO.... I edited the ebuild (into -r1) and through trial and error and headache, determined that the following changes are necessary to get it to work: --- libmcrypt-2.5.1.ebuild 2002-06-17 10:19:59.000000000 -0500 +++ libmcrypt-2.5.1-r1.ebuild 2002-06-20 15:17:55.000000000 -0500 @@ -18,6 +18,8 @@ # Doesn't work with --host bug #3517 # econf --host=${CHOST} \ + export WANT_AUTOMAKE_1_6=1 + autoreconf --force --install --symlink || die econf --disable-posix-threads || die # PHP manual states to disable posix threads, no further explanation given, but i'll Also please note that AUTOMAKE-1.6 *should* be the default whenever 1.5 could be used -- it's actually less buggy and scary and "more" compatible, but I'm sure that's not your department. ;-)
Patch applied. I use automake and the other autotools in my C projects, but every version is the same to me :) kidding. How is it going with Boa ? :)
Boa is just peachy. Thanks for asking! Look for a new release RSN.
the configure fails for libltdl not libmcrypt. Anyway, as for the libtool fixes you've proposed, that created a whole host of other warnings. Wasn't pretty at all with the output. I've implemented an alternate fix -- would you please test that? It's masked..
First, it's not masked! Second, it doesn't work anymore. As for the warnings, that shouldn't matter so long as warnings != errors. ;-) configure: running /bin/sh './configure' --prefix=/usr '--prefix=/usr' '--mandir=/usr/share/man' '--infodir=/usr/share/info' '--datadir=/usr/share' '--sysconfdir=/etc' '--localstatedir=/var/lib' '--disable-posix-threads' 'CFLAGS=-mcpu=i686 -O3 -pipe -g' 'LDFLAGS=-g' --enable-ltdl-install=no --cache-file=/dev/null --srcdir=. configure: warning: CFLAGS=-mcpu=i686 -O3 -pipe -g: invalid host type configure: warning: LDFLAGS=-g: invalid host type configure: error: can only configure for one host and one target at a time configure: error: /bin/sh './configure' failed for libltdl !!! ERROR: The ebuild did not complete successfully. !!! Function src_compile, Line -107, Exitcode 1 !!! (no error message) !!! emerge aborting on /usr/portage/dev-libs/libmcrypt/libmcrypt-2.5.1-r2.ebuild .
check -r2
Um, I *did*. Pasting from my prior message: !!! ERROR: The ebuild did not complete successfully. !!! Function src_compile, Line -107, Exitcode 1 !!! (no error message) !!! emerge aborting on /usr/portage/dev-libs/libmcrypt/libmcrypt-2.5.1-r2.ebuild . Still not fixed.
What's your CHOST line? libtool doesn't do anything but i[3456]86. Looks to me like that's what the complaint is...
CHOST=i686-pc-linux-gnu The problem is *not* the CHOST variable. It's the broken automake that was used when the package was originally built by upstream. I really fail to see what's wrong with the patch I supplied - it fixes the (underlying) problem and despite the "warnings" generated it is still a good fix. There might be a "better" solution out there, but not one that is likely to work all the time.
Az, Please have a look at this? Thanks :)
Yo Jon, can you maybe just try -r2 again after a rsync or cvs update ? I updated our libtool stuff a day or two ago to fix a problem where some deps was dropped. Also maybe update libtool beforehand.
I am completely up-to-date as of 30 June 2002, roughly noon (CST). I still get: configure: running /bin/sh './configure' --prefix=/usr '--prefix=/usr' '--mandir=/usr/share/man' '--infodir=/usr/share/info' '--datadir=/usr/share' '--sysconfdir=/etc' '--localstatedir=/var/lib' '--disable-posix-threads' 'CFLAGS=-mcpu=i686 -O3 -pipe -g' 'LDFLAGS=-g' --enable-ltdl-install=no --cache-file=/dev/null --srcdir=. and compilation fails. Still no go. Can we just please fix this problem and move on?
Hrm, to be honest, I dont have a problem fixing it. The problem is that I dont have your problem, so fixing without knowing what to fix is a bit of a problem.
OK, try -r3, maybe just regeneration everything will do it.
-r3 works for me. Yay! There are some minor optimizations that could be made, but it looks good! Note that my solution probably wouldn't be so noisy if WANT_AUTOCONF_2_5=1 was set also, and while I feel it's "cleaner" (and more correct), this works too. Technically one *should* use autoreconf whenever possible, because it takes into account when to use aclocal -I. and when not to, etc... Perhaps using *both* export WANT_AUTOMAKE_1_6=1 export WANT_AUTOCONF_2_5=1 then autoreconf --force --install --symlink || die might work. However, it's good to go for me!
Ok, great stuff. Just to check, what CVS revision is that ebuild .. I must confess I made changes afterwards :P
Anyhow, I did the other changes, and it works fine here. The autoreconf part should look as follow, so please update -r3 and test: # Try to fix some wierd build problems. See bug #3940. echo ">>> Reconfiguring package..." for x in $(find ${S} -name configure) do cd ${x%/*} export WANT_AUTOMAKE_1_6=1 export WANT_AUTOCONF_2_5=1 autoreconf --force --install --symlink &>${T}/autoreconf.log || ( \ echo "DEBUG: working directory is: `pwd`" >>${T}/autoreconf.log eerror "Reonfigure failed, please attatch the contents of:" eerror eerror " ${T}/autoreconf.log" eerror eerror "in your bugreport." # we need an error here, else the ebuild do not die exit 1 ) || die "running autoreconf failed" done
revision 1.3
OK, one final note. I updated this morning (Mon Jul 1 13:30 UTC 2002), and optimizied/fixed -r3 and turned it into -r4. The changes below are useful because *autoreconf* (note the 're'), automatically recurses. Other than that, the only "real" change is indenting. See below for "real" patch and patch with diff told to ignore changes in whitespace. --- libmcrypt-2.5.1-r3.ebuild 2002-07-01 08:36:15.000000000 -0500 +++ libmcrypt-2.5.1-r4.ebuild 2002-07-01 08:45:14.000000000 -0500 @@ -29,22 +29,18 @@ # Try to fix some wierd build problems. See bug #3940. echo ">>> Reconfiguring package..." - for x in $(find ${S} -name configure) - do - cd ${x%/*} - export WANT_AUTOMAKE_1_6=1 - export WANT_AUTOCONF_2_5=1 - autoreconf --force --install --symlink &>${T}/autoreconf.log || ( \ - echo "DEBUG: working directory is: `pwd`" >>${T}/autoreconf.log - eerror "Reonfigure failed, please attatch the contents of:" - eerror - eerror " ${T}/autoreconf.log" - eerror - eerror "in your bugreport." - # we need an error here, else the ebuild do not die - exit 1 - ) || die "running autoreconf failed" - done + export WANT_AUTOMAKE_1_6=1 + export WANT_AUTOCONF_2_5=1 + autoreconf --force --install --symlink &>${T}/autoreconf.log || ( \ + echo "DEBUG: working directory is: `pwd`" >>${T}/autoreconf.log + eerror "Reonfigure failed, please attatch the contents of:" + eerror + eerror " ${T}/autoreconf.log" + eerror + eerror "in your bugreport." + # we need an error here, else the ebuild do not die + exit 1 + ) || die "running autoreconf failed" } src_compile() { --- libmcrypt-2.5.1-r3.ebuild 2002-07-01 08:36:15.000000000 -0500 +++ libmcrypt-2.5.1-r4.ebuild 2002-07-01 08:45:14.000000000 -0500 @@ -29,9 +29,6 @@ # Try to fix some wierd build problems. See bug #3940. echo ">>> Reconfiguring package..." - for x in $(find ${S} -name configure) - do - cd ${x%/*} export WANT_AUTOMAKE_1_6=1 export WANT_AUTOCONF_2_5=1 autoreconf --force --install --symlink &>${T}/autoreconf.log || ( \ @@ -44,7 +41,6 @@ # we need an error here, else the ebuild do not die exit 1 ) || die "running autoreconf failed" - done } src_compile() {
Thanks Jon, guess I still have a lot to learn with the auto* sucko, but with help well get there :) Anyhow, I am commiting -r4, and closing this bug.
Has it actually been committed yet? I just checked it out and no -r4 as yet. This is -500 13:14