[Buildroot] _VER vs. _VERSION breakage

[Buildroot] _VER vs. _VERSION breakage

[Bernhard Fischer]

Ulf,

Your change to use _VERSION broke at least acpid. Please double-check
that all packages that you touched in this patch still build fine and
repair.

thanks,

[Buildroot] _VER vs. _VERSION breakage

[Ulf Samuelsson]

s?n 2007-07-15 klockan 12:40 +0200 skrev Bernhard Fischer:
> Ulf,
> 
> Your change to use _VERSION broke at least acpid. Please double-check
> that all packages that you touched in this patch still build fine and
> repair.

Here is the result of trying to build most available packages
in buildroot for ARM. (I build for ARM generic (not using WCHAR)).

A lot of the packages does not build for ARM,
but if you can download the package, decompress it
in build_<arch> and then configure or patch it, 
then the VERSION patch is OK for this package.

At least some of the packages seems to require the uClibc is 
built with WCHAR. We should probably identify these packages
and hide them during configuration if WCHAR is not enabled.

Some packages like ACPID does not build for ARM
but I think I got that right anyway, even if I did not try that
specific package.
Maybe someone building x86 can report the few packages
I have problems with on ARM?


acpid/acpid.mk			- *******************
*argus/argus.mk			- OK
*at/at.mk			- OK
*autoconf/autoconf.mk		- OK
*automake/automake.mk		- OK
avahi/avahi.mk			- *******************
				- Requires "libdaemon" which fails due 
				  to AC_PROG_LIBTOOL not recognized
*bash/bash.mk			- OK
*berkeleydb/berkeleydb.mk	- OK
*bind/bind.mk			- OK
*bison/bison.mk			- OK
*bzip2/bzip2.mk			- OK
*coreutils/coreutils.m		- OK (Fails to build on ARM)
*cvs/cvs.mk			- OK (Fails to build on ARM)
*dash/dash.mk			- OK
*dbus/dbus.mk			- OK (Fails to build on ARM)
dbus-glib/dbus-glib.mk		- *************************
				- Requires "dbus" which fails on ARM
*dhcp/dhcp.mk			- OK
*diffutils/diffutils.mk		- OK
*distcc/distcc.mk		- OK
*dmalloc/dmalloc.mk		- OK
*dnsmasq/dnsmasq.mk		- OK
*dropbear/dropbear.mk		- OK
*e2fsprogs/e2fsprogs.mk		- OK
*file/file.mk			- OK
*findutils/findutils.mk		- OK (Fails to build on ARM)
*gawk/gawk.mk			- OK (Fails to build on ARM)
*gettext/gettext.mk		- OK (Fails to build on ARM)
*grep/grep.mk			- OK
*gzip/gzip.mk			- OK
hal/hal.mk			- *************************
				- Requires "dbus" which fails on ARM
*hotplug/hotplug.mk		- OK
*iostat/iostat.mk		- OK
*iproute2/iproute2.mk		- OK
*ipsec-tools/ipsec-tools.mk	- OK
*iptables/iptables.mk		- OK
*jpeg/jpeg.mk			- OK
*kexec/kexec.mk			- OK (ARM = unsupported architecture)
*l2tp/l2tp.mk			- OK
				  Patch version 2.1 has disappeared, 
				  2.2 is current. Fails to build
*libdaemon/libdaemon.mk		- OK (Fails to build on ARM)
*libelf/libelf.mk		- OK
*libevent/libevent.mk		- OK
libglib12/libglib12.mk		- *************************
libgtk12/libgtk12.mk		- *************************
*libpcap/libpcap.mk		- OK
*libpng/libpng.mk		- OK
*libsndfile/libsndfile.mk	- OK
libsysfs/libsysfs.mk		- OK
*libtool/libtool.mk		- OK
*libusb/libusb.mk		- OK
*lighttpd/lighttpd.mk		- init.d script changed name/directory.
*links/links.mk			- OK (Fails to build on ARM)
				  Needs -I$(STAGING_DIR)/include
*lrzsz/lrzsz.mk			- OK
*ltt/ltt.mk			- OK
*lzma/lzma.mk			- OK
*m4/m4.mk			- OK
*make/make.mk			- OK 
				Make fails during Cross-Compiles since
				it calls the Cross Compiled Make binary 
				in the latter part of the build instead 
				of the /usr/bin/make
*memtester/memtester.mk		- OK
metacity/metacity.mk		- *************************
*microcom/microcom.mk		- OK
*microperl/microperl.mk		- OK
*mkdosfs/mkdosfs.mk		- OK
*modutils/modutils.mk		- OK
*mpatrol/mpatrol.mk		- OK (Fails to build on ARM)
*mrouted/mrouted.mk		- OK (Fails to build on ARM)
*nano/nano.mk			- OK
*ncurses/ncurses.mk		- OK
*netkitbase/netkitbase.mk	- OK (Fails to build on ARM)
netkittelnet/netkittelnet.mk	- *************************
*netplug/netplug.mk		- OK (Requires bk (bitkeeper))
*netsnmp/netsnmp.mk		- OK Needs "defined(xxx) 
*nfs-utils/nfs-utils.mk		- OK (Fails to build on ARM)
*openssh/openssh.mk		- OK (Fails to build on ARM)
				  Needs to include $(STAGING_DIR)/ 
				  include
				  64/64 division
*openssl/openssl.mk		- OK
*patch/patch.mk			- OK
*pciutils/pciutils.mk		- OK, requires inux
*pkgconfig/pkgconfig.mk		- OK
*portmap/portmap.mk		- OK (Fails to build on ARM)
*procps/procps.mk		- OK (Fails to build on ARM)
				  wchar.h => bits/wchar.h,wctype.h 
					-I???
*proftpd/proftpd.mk		- OK
*psmisc/psmisc.mk		- OK (Fails to build on ARM)
qtopia4/qtopia4.mk		- *************************
*quagga/quagga.mk		- OK (Fails to build on ARM) (no SNMP)
raidtools/raidtools.mk		- *************************
*readline/readline.mk		- OK
*rsync/rsync.mk			- OK
*ruby/ruby.mk			- OK (Fails to build on ARM) 
					  (tried 1.8.2,1.86 + svn)
*samba/samba.mk			- OK (Fails to build on ARM)
					(tries to run test program
					compiled for ARM 
					during configure)
*sed/sed.mk			- OK (Fails to build on ARM)
					(WCHAR)
*sfdisk/sfdisk.mk		- OK (Fails to build on ARM)
*slang/slang.mk			- OK
*smartmontools/smartmontools.mk	- OK
*strace/strace.mk		- OK
*stunnel/stunnel.mk		- OK (Fails to build on ARM)
					needs SSL
*sudo/sudo.mk			- OK
*sysklogd/sysklogd.mk		- OK (Fails to build on ARM)
				 	lacks "linux/linkage.h"
*sysvinit/sysvinit.mk		- OK
*tar/tar.mk			- OK (Fails to build on ARM)
					(wcint_t not declared)
*tcpdump/tcpdump.mk		- OK
*tftpd/tftpd.mk			- OK
*thttpd/thttpd.mk		- OK
*tiff/tiff.mk			- OK
*tinylogin/tinylogin.mk		- OK
tinyx/tinyx.mk			- OK (Fails to build on ARM)
					cannot find -lXft
tn5250/tn5250.mk		- OK (Fails to build on ARM)
					needs slang.h
ttcp/ttcp.mk			- OK (Fails to build on ARM)
					needs #include <strings.h>
*udhcp/udhcp.mk			- This was merged into busybox!
*uemacs/uemacs.mk		- OK
*usbutils/usbutils.mk		- OK
*util-linux/util-linux.mk	- OK (Fails to build on ARM)
*wget/wget.mk			- OK
*which/which.mk			- OK
*wipe/wipe.mk			- OK
*wireless-tools/wireless-tools.mk OK
*vtun/vtun.mk			- OK
*xfsprogs/xfsprogs.mk		- OK (Fails to build on ARM)
*zlib/zlib.mk			- OK
=====================
Some other failures (not affected by VERSION patch)
asterisk
DirectFB
dmraid
fontconfig
libdaemon/libdaemon.mk Fails.
	avahi depends on this.
curl does not download	7.13.1 is too old...
	7.16.4 cannot be patched...
libid3tag does not compile
	=> madplay does not compile
lvm2



> 
> thanks,
> _______________________________________________
> buildroot mailing list
> buildroot at uclibc.org
> http://busybox.net/mailman/listinfo/buildroot

[Buildroot] _VER vs. _VERSION breakage

[Bernhard Fischer]

On Mon, Jul 16, 2007 at 09:54:06PM +0200, Ulf Samuelsson wrote:
>s?n 2007-07-15 klockan 12:40 +0200 skrev Bernhard Fischer:
>> Ulf,
>> 
>> Your change to use _VERSION broke at least acpid. Please double-check
>> that all packages that you touched in this patch still build fine and
>> repair.
>
>Here is the result of trying to build most available packages
>in buildroot for ARM. (I build for ARM generic (not using WCHAR)).

I will revert that _VERSION vs. _VER patch since it breaks a couple of
packages.
>
>A lot of the packages does not build for ARM,

Could be. For changes like _VER vs. _VERSION, a quick check with an x86
build before and after such a change would have been the proper thing to
do, imho.

>but if you can download the package, decompress it
>in build_<arch> and then configure or patch it, 
>then the VERSION patch is OK for this package.

>At least some of the packages seems to require the uClibc is 
>built with WCHAR. We should probably identify these packages
>and hide them during configuration if WCHAR is not enabled.

Yes, this is a completely different issue, though.

>
>Some packages like ACPID does not build for ARM
>but I think I got that right anyway, even if I did not try that
>specific package.
>Maybe someone building x86 can report the few packages
>I have problems with on ARM?

No need, i'll revert said patch and let those who want such a patch to
go in check if they break something.

Breaking the build for cosmetic reasons is not a sensible thing to do.

[Buildroot] _VER vs. _VERSION breakage

[Ulf Samuelsson]

>>> Your change to use _VERSION broke at least acpid. Please double-check
>>> that all packages that you touched in this patch still build fine and
>>> repair.
>>
>>Here is the result of trying to build most available packages
>>in buildroot for ARM. (I build for ARM generic (not using WCHAR)).
> 
> I will revert that _VERSION vs. _VER patch since it breaks a couple of
> packages.



>>A lot of the packages does not build for ARM,
> 
> Could be. For changes like _VER vs. _VERSION, a quick check with an x86
> build before and after such a change would have been the proper thing to
> do, imho.
> 
>>but if you can download the package, decompress it
>>in build_<arch> and then configure or patch it, 
>>then the VERSION patch is OK for this package.
> 
>>At least some of the packages seems to require the uClibc is 
>>built with WCHAR. We should probably identify these packages
>>and hide them during configuration if WCHAR is not enabled.
> 
> Yes, this is a completely different issue, though.
> 
>>
>>Some packages like ACPID does not build for ARM
>>but I think I got that right anyway, even if I did not try that
>>specific package.
>>Maybe someone building x86 can report the few packages
>>I have problems with on ARM?
> 
> No need, i'll revert said patch and let those who want such a patch to
> go in check if they break something.
> 
> Breaking the build for cosmetic reasons is not a sensible thing to do.
>

It is not only for cosmetic things alone.

Instead of reverting the patch, I can check on x86 instead today and fix if neccessary.
Please tell me which packages are broken.
acpid wass, but that should be fixed by now.


Best Regards
Ulf Samuelsson

[Buildroot] _VER vs. _VERSION breakage

[Cristian Ionescu-Idbohrn]

On Mon, 23 Jul 2007, Bernhard Fischer wrote:

> On Mon, Jul 16, 2007 at 09:54:06PM +0200, Ulf Samuelsson wrote:
> >s?n 2007-07-15 klockan 12:40 +0200 skrev Bernhard Fischer:
> >> Ulf,
> >>
> >> Your change to use _VERSION broke at least acpid. Please double-check
> >> that all packages that you touched in this patch still build fine and
> >> repair.
> >
> >Here is the result of trying to build most available packages
> >in buildroot for ARM. (I build for ARM generic (not using WCHAR)).
>
> I will revert that _VERSION vs. _VER patch since it breaks a couple of
> packages.

Now this is IMO a rather silly :( You made your point.  Can't you just
fix those "couple of packages" instead and stop this ping-pong game?


Cheers,

-- 
Cristian

[Buildroot] _VER vs. _VERSION breakage

[Bernhard Fischer]

On Mon, Jul 23, 2007 at 12:22:41PM +0200, Cristian Ionescu-Idbohrn wrote:
>On Mon, 23 Jul 2007, Bernhard Fischer wrote:
>
>> On Mon, Jul 16, 2007 at 09:54:06PM +0200, Ulf Samuelsson wrote:
>> >s?n 2007-07-15 klockan 12:40 +0200 skrev Bernhard Fischer:
>> >> Ulf,
>> >>
>> >> Your change to use _VERSION broke at least acpid. Please double-check
>> >> that all packages that you touched in this patch still build fine and
>> >> repair.
>> >
>> >Here is the result of trying to build most available packages
>> >in buildroot for ARM. (I build for ARM generic (not using WCHAR)).
>>
>> I will revert that _VERSION vs. _VER patch since it breaks a couple of
>> packages.
>
>Now this is IMO a rather silly :( You made your point.  Can't you just
>fix those "couple of packages" instead and stop this ping-pong game?

I don't have time to fix them and since they worked before, reverting it
is fine. It's not that i didn't wait for 1 week (!) to allow Ulf (or
anybody else) to fix it up.. Selecting i386 in the config isn't really
hard, i suppose.
pong.

[Buildroot] _VER vs. _VERSION breakage

[Ulf Samuelsson]

m?n 2007-07-23 klockan 13:04 +0200 skrev Bernhard Fischer:
> On Mon, Jul 23, 2007 at 12:22:41PM +0200, Cristian Ionescu-Idbohrn wrote:
> >On Mon, 23 Jul 2007, Bernhard Fischer wrote:
> >
> >> On Mon, Jul 16, 2007 at 09:54:06PM +0200, Ulf Samuelsson wrote:
> >> >s?n 2007-07-15 klockan 12:40 +0200 skrev Bernhard Fischer:
> >> >> Ulf,
> >> >>
> >> >> Your change to use _VERSION broke at least acpid. Please double-check
> >> >> that all packages that you touched in this patch still build fine and
> >> >> repair.
> >> >
> >> >Here is the result of trying to build most available packages
> >> >in buildroot for ARM. (I build for ARM generic (not using WCHAR)).
> >>
> >> I will revert that _VERSION vs. _VER patch since it breaks a couple of
> >> packages.
> >
> >Now this is IMO a rather silly :( You made your point.  Can't you just
> >fix those "couple of packages" instead and stop this ping-pong game?
> 
> I don't have time to fix them and since they worked before, reverting it
> is fine. It's not that i didn't wait for 1 week (!) to allow Ulf (or
> anybody else) to fix it up.. Selecting i386 in the config isn't really
> hard, i suppose.

Have the machine working on it at the moment, testing x86...
Bear with me.


> pong.
> _______________________________________________
> buildroot mailing list
> buildroot at uclibc.org
> http://busybox.net/mailman/listinfo/buildroot