Re: [oe-commits] Otavio Salvador : systemd.bbclass: depends on systemd

Re: [oe-commits] Otavio Salvador : systemd.bbclass: depends on systemd

[Martin Jansa]

[-- Attachment #1: Type: text/plain, Size: 2019 bytes --]

On Wed, Feb 01, 2012 at 08:04:41AM +0000, git@git.openembedded.org wrote:
> Module: meta-openembedded.git
> Branch: master
> Commit: c029d4a268de1ac7076fbeb9e337e96fdb1b226d
> URL:    http://git.openembedded.org/?p=meta-openembedded.git&a=commit;h=c029d4a268de1ac7076fbeb9e337e96fdb1b226d
> 
> Author: Otavio Salvador <otavio@ossystems.com.br>
> Date:   Tue Jan 31 17:38:57 2012 +0000
> 
> systemd.bbclass: depends on systemd
> 
> Packages that uses the systemd class need to have systemd build
> otherwise they might fail to detect systemd as available or fail to
> generate rootfs.

Can we move this only to packages which are really depending on systemd
compiled?

I have use-case where
small cpio image has busybox
busybox depends on systemd just because of this
systemd depends on kernel (because RRECOMMENDS on modules)
kernel depends on that cpio image which is used as ramdisk

So in order to finish cpio.do_rootfs I have to finish kernel and kernel
fails to build because doesn't have cpio available yet (DEPENDS = "cpio"
is not enough because doesn't need do_rootfs to finish AFAIK).

Cheers,

> 
> Signed-off-by: Otavio Salvador <otavio@ossystems.com.br>
> Signed-off-by: Koen Kooi <koen@dominion.thruhere.net>
> 
> ---
> 
>  meta-oe/classes/systemd.bbclass |    2 +-
>  1 files changed, 1 insertions(+), 1 deletions(-)
> 
> diff --git a/meta-oe/classes/systemd.bbclass b/meta-oe/classes/systemd.bbclass
> index 2ca7bfc..094a12c 100644
> --- a/meta-oe/classes/systemd.bbclass
> +++ b/meta-oe/classes/systemd.bbclass
> @@ -1,4 +1,4 @@
> -DEPENDS_append = " systemd-systemctl-native"
> +DEPENDS_append = " systemd systemd-systemctl-native"
>  
>  systemd_postinst() {
>  OPTS=""
> 
> 
> _______________________________________________
> Openembedded-commits mailing list
> Openembedded-commits@lists.openembedded.org
> http://lists.linuxtogo.org/cgi-bin/mailman/listinfo/openembedded-commits

-- 
Martin 'JaMa' Jansa     jabber: Martin.Jansa@gmail.com

[-- Attachment #2: Digital signature --]
[-- Type: application/pgp-signature, Size: 205 bytes --]

Re: [oe-commits] Otavio Salvador : systemd.bbclass: depends on systemd

[Otavio Salvador]

On Fri, Feb 3, 2012 at 11:14, Martin Jansa <martin.jansa@gmail.com> wrote:

> Can we move this only to packages which are really depending on systemd
> compiled?
>
> I have use-case where
> small cpio image has busybox
> busybox depends on systemd just because of this
> systemd depends on kernel (because RRECOMMENDS on modules)
> kernel depends on that cpio image which is used as ramdisk
>
> So in order to finish cpio.do_rootfs I have to finish kernel and kernel
> fails to build because doesn't have cpio available yet (DEPENDS = "cpio"
> is not enough because doesn't need do_rootfs to finish AFAIK).
>

I understand your use case but it seems in this case you ought to override
it in a bbappend. It doesn't seems right as this is the exception, not the
rule. If we don't depends on it, we can end with the binaries being not
available in case a rootfs generation.

-- 
Otavio Salvador                             O.S. Systems
E-mail: otavio@ossystems.com.br  http://www.ossystems.com.br
Mobile: +55 53 9981-7854              http://projetos.ossystems.com.br

Re: [oe-commits] Otavio Salvador : systemd.bbclass: depends on systemd

[Martin Jansa]

[-- Attachment #1: Type: text/plain, Size: 1154 bytes --]

On Fri, Feb 03, 2012 at 01:30:37PM -0200, Otavio Salvador wrote:
> On Fri, Feb 3, 2012 at 11:14, Martin Jansa <martin.jansa@gmail.com> wrote:
> 
> > Can we move this only to packages which are really depending on systemd
> > compiled?
> >
> > I have use-case where
> > small cpio image has busybox
> > busybox depends on systemd just because of this
> > systemd depends on kernel (because RRECOMMENDS on modules)
> > kernel depends on that cpio image which is used as ramdisk
> >
> > So in order to finish cpio.do_rootfs I have to finish kernel and kernel
> > fails to build because doesn't have cpio available yet (DEPENDS = "cpio"
> > is not enough because doesn't need do_rootfs to finish AFAIK).
> >
> 
> I understand your use case but it seems in this case you ought to override
> it in a bbappend. It doesn't seems right as this is the exception, not the
> rule. If we don't depends on it, we can end with the binaries being not
> available in case a rootfs generation.

Which recipes really depend on systemd during build? I know about
polkit, something else?

-- 
Martin 'JaMa' Jansa     jabber: Martin.Jansa@gmail.com

[-- Attachment #2: Digital signature --]
[-- Type: application/pgp-signature, Size: 205 bytes --]

Re: [oe-commits] Otavio Salvador : systemd.bbclass: depends on systemd

[Otavio Salvador]

On Fri, Feb 3, 2012 at 13:47, Martin Jansa <martin.jansa@gmail.com> wrote:

> > I understand your use case but it seems in this case you ought to
> override
> > it in a bbappend. It doesn't seems right as this is the exception, not
> the
> > rule. If we don't depends on it, we can end with the binaries being not
> > available in case a rootfs generation.
>
> Which recipes really depend on systemd during build? I know about
> polkit, something else?


At this moment I know about polkit, I didn't check newer udev but I think
it ought to depends on it as well.

-- 
Otavio Salvador                             O.S. Systems
E-mail: otavio@ossystems.com.br  http://www.ossystems.com.br
Mobile: +55 53 9981-7854              http://projetos.ossystems.com.br

Re: [oe-commits] Otavio Salvador : systemd.bbclass: depends on systemd

[Andrea Adami]

On Fri, Feb 3, 2012 at 5:58 PM, Otavio Salvador <otavio@ossystems.com.br> wrote:
> On Fri, Feb 3, 2012 at 13:47, Martin Jansa <martin.jansa@gmail.com> wrote:
>
>> > I understand your use case but it seems in this case you ought to
>> override
>> > it in a bbappend. It doesn't seems right as this is the exception, not
>> the
>> > rule. If we don't depends on it, we can end with the binaries being not
>> > available in case a rootfs generation.
>>
>> Which recipes really depend on systemd during build? I know about
>> polkit, something else?
>
>
> At this moment I know about polkit, I didn't check newer udev but I think
> it ought to depends on it as well.
>
> --
> Otavio Salvador                             O.S. Systems
> E-mail: otavio@ossystems.com.br  http://www.ossystems.com.br
> Mobile: +55 53 9981-7854              http://projetos.ossystems.com.br
> _______________________________________________
> Openembedded-devel mailing list
> Openembedded-devel@lists.openembedded.org
> http://lists.linuxtogo.org/cgi-bin/mailman/listinfo/openembedded-devel

Unfortunately the patch has a bad side-effect:

http://paste.debian.net/154741/

(I removed /systemd to show the culprit)

Sadly enough freedesktop git is down, so I've been obliged to revert that patch.

Andrea

Re: [oe-commits] Otavio Salvador : systemd.bbclass: depends on systemd

[Koen Kooi]

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

Op 04-02-12 01:15, Andrea Adami schreef:
> On Fri, Feb 3, 2012 at 5:58 PM, Otavio Salvador <otavio@ossystems.com.br>
> wrote:
>> On Fri, Feb 3, 2012 at 13:47, Martin Jansa <martin.jansa@gmail.com>
>> wrote:
>> 
>>>> I understand your use case but it seems in this case you ought to
>>> override
>>>> it in a bbappend. It doesn't seems right as this is the exception,
>>>> not
>>> the
>>>> rule. If we don't depends on it, we can end with the binaries being
>>>> not available in case a rootfs generation.
>>> 
>>> Which recipes really depend on systemd during build? I know about 
>>> polkit, something else?
>> 
>> 
>> At this moment I know about polkit, I didn't check newer udev but I
>> think it ought to depends on it as well.
>> 
>> -- Otavio Salvador                             O.S. Systems E-mail:
>> otavio@ossystems.com.br  http://www.ossystems.com.br Mobile: +55 53
>> 9981-7854              http://projetos.ossystems.com.br 
>> _______________________________________________ Openembedded-devel
>> mailing list Openembedded-devel@lists.openembedded.org 
>> http://lists.linuxtogo.org/cgi-bin/mailman/listinfo/openembedded-devel
> 
> Unfortunately the patch has a bad side-effect:
> 
> http://paste.debian.net/154741/
> 
> (I removed /systemd to show the culprit)
> 
> Sadly enough freedesktop git is down, so I've been obliged to revert that
> patch.

only cgit, not the actual git server
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.11 (Darwin)
Comment: GPGTools - http://gpgtools.org

iEYEARECAAYFAk8sfH8ACgkQMkyGM64RGpFOKACgk2/B+ac8q6HOs2ctoJff3KrN
D7cAn2w+6lseVIR15bhtOywS9gUvdUFd
=2BCI
-----END PGP SIGNATURE-----

Re: [oe-commits] Otavio Salvador : systemd.bbclass: depends on systemd

[Andreas Müller]

>> Which recipes really depend on systemd during build? I know about
>> polkit, something else?

udisks [1]

[1] http://lists.linuxtogo.org/pipermail/openembedded-devel/2012-January/037230.html

Andreas

Re: [oe-commits] Otavio Salvador : systemd.bbclass: depends on systemd

[Peter Bigot]

On Fri, Feb 3, 2012 at 8:00 PM, Andreas Müller
<schnitzeltony@googlemail.com> wrote:
>>> Which recipes really depend on systemd during build? I know about
>>> polkit, something else?
>
> udisks [1]

Yeah, but that might be a consequence of udisks depending on polkit.
It was a libsystemd-login reference in a polkit .la file dependency
that made me install that patch so I could get udisks to build.

>
> [1] http://lists.linuxtogo.org/pipermail/openembedded-devel/2012-January/037230.html
>
> Andreas
>
> _______________________________________________
> Openembedded-devel mailing list
> Openembedded-devel@lists.openembedded.org
> http://lists.linuxtogo.org/cgi-bin/mailman/listinfo/openembedded-devel

Re: [oe-commits] Otavio Salvador : systemd.bbclass: depends on systemd

[Khem Raj]

On (04/02/12 03:00), Andreas Müller wrote:
> >> Which recipes really depend on systemd during build? I know about
> >> polkit, something else?

why not make it a globally selectable ?
so we can make sure that autoconf does not try to guess
but we tell it if systemd features should {en|dis}abled
explicitly and something like virtual/init-manager or something
since I think it has to coexist with sysvinit or upstart
whatever one chooses.

Re: [oe-commits] Otavio Salvador : systemd.bbclass: depends on systemd

[Khem Raj]

On (03/02/12 20:15), Peter Bigot wrote:
> On Fri, Feb 3, 2012 at 8:00 PM, Andreas Müller
> <schnitzeltony@googlemail.com> wrote:
> >>> Which recipes really depend on systemd during build? I know about
> >>> polkit, something else?
> >
> > udisks [1]
> 
> Yeah, but that might be a consequence of udisks depending on polkit.
> It was a libsystemd-login reference in a polkit .la file dependency
> that made me install that patch so I could get udisks to build.


polkit should also become a distro feature FWIW

Re: [oe-commits] Otavio Salvador : systemd.bbclass: depends on systemd

[Joshua Lock]

On 03/02/12 18:18, Khem Raj wrote:
> On (04/02/12 03:00), Andreas Müller wrote:
>>>> Which recipes really depend on systemd during build? I know about
>>>> polkit, something else?
>
> why not make it a globally selectable ?
> so we can make sure that autoconf does not try to guess
> but we tell it if systemd features should {en|dis}abled
> explicitly and something like virtual/init-manager or something
> since I think it has to coexist with sysvinit or upstart
> whatever one chooses.

We should definitely do this, I've not had enough time to figure out. I 
think it's going to have to be a distro level thing as so far as I can 
tell we need to be able to:
o set the init system
o include certain packages
o potentially enable/disable configure flags

I've not had much opportunity to look into this yet but we need to be 
certain whichever mechanism we enable works for systemd, sysvinit, 
upstart, s6, whatever.

Cheers,
Joshua
-- 
Joshua Lock
         Yocto Project "Johannes factotum"
         Intel Open Source Technology Centre

Re: [oe-commits] Otavio Salvador : systemd.bbclass: depends on systemd

[Martin Jansa]

[-- Attachment #1: Type: text/plain, Size: 1332 bytes --]

On Fri, Feb 03, 2012 at 08:15:46PM -0600, Peter Bigot wrote:
> On Fri, Feb 3, 2012 at 8:00 PM, Andreas Müller
> <schnitzeltony@googlemail.com> wrote:
> >>> Which recipes really depend on systemd during build? I know about
> >>> polkit, something else?
> >
> > udisks [1]
> 
> Yeah, but that might be a consequence of udisks depending on polkit.
> It was a libsystemd-login reference in a polkit .la file dependency
> that made me install that patch so I could get udisks to build.

OK, so I have 21 recipes which inherit systemd and 2 need build-time
systemd dependency and the rest is fine with RDEPENDS, can we please 
move systemd from inherited DEPENDS to those 2 recipe (polkit and
udisks)?

Cheers,

> 
> >
> > [1] http://lists.linuxtogo.org/pipermail/openembedded-devel/2012-January/037230.html
> >
> > Andreas
> >
> > _______________________________________________
> > Openembedded-devel mailing list
> > Openembedded-devel@lists.openembedded.org
> > http://lists.linuxtogo.org/cgi-bin/mailman/listinfo/openembedded-devel
> 
> _______________________________________________
> Openembedded-devel mailing list
> Openembedded-devel@lists.openembedded.org
> http://lists.linuxtogo.org/cgi-bin/mailman/listinfo/openembedded-devel

-- 
Martin 'JaMa' Jansa     jabber: Martin.Jansa@gmail.com

[-- Attachment #2: Digital signature --]
[-- Type: application/pgp-signature, Size: 205 bytes --]