Hi,
As development group maintainer I start to see requests (mostly from Cristian Rodriguez) that remove sysvinit, xinetd support completely from packages and also remove syslog dependencies.
What is the current status?
Is the total sysvinit removal and total xinetd removal now fact and the future plan?
Ciao, Marcus
On 31/10/12 09:20, Marcus Meissner wrote:
Hi,
As development group maintainer I start to see requests (mostly from Cristian Rodriguez) that remove sysvinit, xinetd support completely from packages and also remove syslog dependencies.
What is the current status?
Is the total sysvinit removal and total xinetd removal now fact and the future plan?
The functionality of xinetd that is currently used by packages is now provided by systemd, I do not aim at removing it, just not providing packages with xinetd configuration files, so we can remove xinetd from the default installation.
Removing dependencies on syslog allow us to run with the journal only.
On 2012-10-31 09:28:24 -0300, Cristian Rodríguez wrote:
On 31/10/12 09:20, Marcus Meissner wrote:
Hi,
As development group maintainer I start to see requests (mostly from Cristian Rodriguez) that remove sysvinit, xinetd support completely from packages and also remove syslog dependencies.
What is the current status?
Is the total sysvinit removal and total xinetd removal now fact and the future plan?
The functionality of xinetd that is currently used by packages is now provided by systemd, I do not aim at removing it, just not providing packages with xinetd configuration files, so we can remove xinetd from the default installation.
Removing dependencies on syslog allow us to run with the journal only.
That still doesnt mean you can remove the files. what about people who want to keep using sysvinit+xinetd? we just drop their support files because you dont need it anymore?
Funnily ... the rsync package e.g. does not even require xinetd! But as a sysvinit user i get forced to install systemd just for using rsync.
maybe we should rip out that requires so we dont force packages on people?
darix
On 31.10.2012 13:36, Marcus Rueckert wrote:
That still doesnt mean you can remove the files. what about people who want to keep using sysvinit+xinetd? we just drop their support files because you dont need it anymore?
Funnily ... the rsync package e.g. does not even require xinetd! But as a sysvinit user i get forced to install systemd just for using rsync.
maybe we should rip out that requires so we dont force packages on people?
This is what happened in parts - removing e.g. sysvinit(syslog) requires, but in general sysvinit is no longer supported in factory, so there is little point in trying to support it in packages itself.
Greetings, Stephan
Hi,
On Wed, Oct 31, Marcus Rueckert wrote:
That still doesnt mean you can remove the files. what about people who want to keep using sysvinit+xinetd? we just drop their support files because you dont need it anymore?
Do you really expect the systemd guys to care about such users?
Funnily ... the rsync package e.g. does not even require xinetd! But as a sysvinit user i get forced to install systemd just for using rsync.
It is even worse! If you want to use NFS mounts, you are already forced into systemd madness:
Sirius:/etc/init.d # insserv rpcbind insserv: Note: sysvinit service rpcbind is shadowed by systemd rpcbind.service, Forwarding request to '/bin/systemctl --no-reload enable rpcbind.service'. ln -s '/lib/systemd/system/rpcbind.service' '/etc/systemd/system/multi-user.target.wants/rpcbind.service' ln -s '/lib/systemd/system/rpcbind.socket' '/etc/systemd/system/sockets.target.wants/rpcbind.socket' insserv: Forward service request to systemctl returned error status : 512
Sirius:/etc/init.d # rpm -e systemd-44-10.1.1.x86_64 error: Failed dependencies: libsystemd-daemon.so.0()(64bit) is needed by (installed) rpcbind-0.2.0_git201103171419-3.1.3.x86_64 libsystemd-daemon.so.0(LIBSYSTEMD_DAEMON_31)(64bit) is needed by (installed) rpcbind-0.2.0_git201103171419-3.1.3.x86_64 systemd is needed by (installed) systemd-presets-branding-basedonopensuse-12.2-3.7.1.x86_64 systemd is needed by (installed) mkinitrd-2.7.1-62.6.1.x86_64 systemd is needed by (installed) cronie-1.4.8-37.2.3.x86_64 systemd is needed by (installed) rpcbind-0.2.0_git201103171419-3.1.3.x86_64 systemd is needed by (installed) openssh-6.0p1-2.3.3.x86_64 systemd >= 44 is needed by (installed) plymouth-0.8.6.1-1.15.1.x86_64 systemd is needed by (installed) ConsoleKit-0.4.5-12.1.2.x86_64 systemd is needed by (installed) yast2-installation-2.22.10-1.1.4.noarch systemd is needed by (installed) postfix-2.8.11-2.6.1.x86_64
I have been told that "insserv returning error(!) status 512 actually means that it was successfull and might even work", but after the clear statement "you cannot uninstall systemd" I finally gave up.
maybe we should rip out that requires so we dont force packages on people?
Well, right now I'm in the process of switching my private machines to some *BSD variant, since running some wannabe-Windows clone is not interesting for me.
It is ironic that tomorrow it is exactly 20 years that I started using Linux :/
R.I.P.
darix
Hubert Mantel
Le mercredi 31 octobre 2012 à 13:20 +0100, Marcus Meissner a écrit :
Hi,
As development group maintainer I start to see requests (mostly from Cristian Rodriguez) that remove sysvinit, xinetd support completely from packages and also remove syslog dependencies.
What is the current status?
Is the total sysvinit removal and total xinetd removal now fact and the future plan?
To be honest, I would prefer people focus their energy in helping fixing bugs remaining in systemd (yes, there are still some), missing integrations compared to what we had with sysvinit, etc.. rather than just removing sysvinit/xinetd support for packages. This is something we can always do in the future..
On 31/10/12 09:32, Frederic Crozat wrote:
To be honest, I would prefer people focus their energy in helping fixing bugs remaining in systemd (yes, there are still some), missing integrations compared to what we had with sysvinit, etc.. rather than just removing sysvinit/xinetd support for packages. This is something we can always do in the future..
I was hoping that we finally put a stop to the stupid madness of multiple init systems, but apparently people still want to beat the dead horse.
If every single change requires me to participate in endless discussion, rejected requests and request to keep legacy stuff around then I rest my case and I'm out of here.
I do not have time or patience for this insanity.
-----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1
On 10/31/2012 01:48 PM, Cristian Rodríguez wrote:
I do not have time or patience for this insanity.
Insane is only that you remove working stuff meanwhile systemd is still broken. Aim at fixing systemd first! You can pick any of the 56 open bugs which make our lifes harder.
thanks, - -- js suse labs
Hi,
On Wed, Oct 31, Cristian Rodríguez wrote:
I was hoping that we finally put a stop to the stupid madness of multiple init systems, but apparently people still want to beat the dead horse.
Be patient. It is just the last rebelling of some mislead people who enjoyed having a stable, maintainable and debuggable system. They will vanish soon.
If every single change requires me to participate in endless discussion, rejected requests and request to keep legacy stuff around then I rest my case and I'm out of here.
Again: Be patient. Soon you can do whatever you want.
I do not have time or patience for this insanity.
Well said.
Hubert Mantel
On 10/31/2012 08:48 AM, Cristian Rodríguez wrote:
On 31/10/12 09:32, Frederic Crozat wrote:
To be honest, I would prefer people focus their energy in helping fixing bugs remaining in systemd (yes, there are still some), missing integrations compared to what we had with sysvinit, etc.. rather than just removing sysvinit/xinetd support for packages. This is something we can always do in the future..
I was hoping that we finally put a stop to the stupid madness of multiple init systems, but apparently people still want to beat the dead horse.
If every single change requires me to participate in endless discussion, rejected requests and request to keep legacy stuff around then I rest my case and I'm out of here.
Well, you cannot just throw out the "legacy" system if the new system is not up to par yet. According to Frederic's message there is still work to be done.
Taking one step at a time lets other people follow easier.
Robert
On Wed, 31 Oct 2012 09:48:10 -0300, Cristian Rodríguez crrodriguez@opensuse.org wrote:
I was hoping that we finally put a stop to the stupid madness of multiple init systems, but apparently people still want to beat the dead horse.
No, but you can only replace a vital part of the system when the replacement fully works. The sooner the remaining bugs in systemd are squashed, the sooner you can remove systemvinit. It's that simple.
Philipp
Hi,
On Wed, Oct 31, Frederic Crozat wrote:
To be honest, I would prefer people focus their energy in helping fixing bugs remaining in systemd (yes, there are still some), missing
Bruhahahaha! ROTFL :)
integrations compared to what we had with sysvinit, etc.. rather than just removing sysvinit/xinetd support for packages. This is something we can always do in the future..
Sure. But why not right now?
Frederic Crozat fcrozat@suse.com SUSE
Hubert Mantel SUSE
Le mercredi 31 octobre 2012 à 14:40 +0100, Hubert Mantel a écrit :
Hi,
On Wed, Oct 31, Frederic Crozat wrote:
To be honest, I would prefer people focus their energy in helping fixing bugs remaining in systemd (yes, there are still some), missing
Bruhahahaha! ROTFL :)
Great, I'll happily review any patch you'll write in the future for fixing bugs reported in systemd.
EOT.