#803 Add johannbg to provenpackager explicitly to work on sysv2systemd conversion
Closed None Opened 12 years ago by toshio.

Viking_Ice has been working on understanding and actually performing conversion from sysv init script to systemd unit files. However, much of the work that he has done is stuck in bugzilla as the maintainers of the packages haven't updated the packages to use the new files. This is somewhat understandable as they don't have the understanding of the new init system to be able to review the new systemd unit files and know that they are going to do what they should. However, this is preventing us from moving forward on this.

Having the person who is driving this forward be able to commit those changes to the packages directly would be a great help here. johannbg (Viking_Ice on IRC) could be made a provenpackager (he would also need to be sponsored into packager) for the explicit purpose of working on the conversion. FESCo could state that they're explicit expectations as well.

Viking_Ice has said: "I'm not a provenpackager and have no interested in becoming a package maintainer all thou I would not mind have the ability to fix things here and there if I came across them." which I don't believe would conflict in spirit with this.

== Proposal ==

  • Sponsor johannbg into packager and provenpackager for the explicit purpose of working on sysV to systemd updates.
  • johannbg can make updates to packages to enable the new scripts just as any other provenpackager can
  • Due to lack of other packaging experience, it's expected that johannbg would collaborate with package maintainers on any changes that don't have to do with the sysv to systemd update.
  • The period between Fn branch and F(n+1) alpha (we're currently in this phase for F18) is ideal for making these changes. Any provenpackager in general and johannbg in specific has explicit permission to apply the changes to migrate packages to systemd during this window.
  • When systemd migration is complete, johannbg's provenpackager status may but is not required to be removed at FESCo and johannbg's choosing.

Usually I want to be given a bit of heads up before someone request something for me ( not that I mind this particular gesture).

I personally prefer not to receive any "special" treatment over any other community members as in I should be treated equally to others.

That said I re-emphasize the point that I have said before on previous occasion that blindly packaging unit and shipping them might not be the best thing to do however the argument can be made that fixing systemd unit is relatively easy so the biggest (proven) challenge is getting those unit shipped in the first place.

Now bear in mind that the absolutely best thing to do here is for maintainers to package and ship the unit's I've migrated thus far.

At least maintainers should provide some kind of feedback/comment on the relevant package at total minimum ack the units for packaging.

The second best thing to do is to have proven packager to package those units.

The above would allow me to continue to focus on migrate the legacy sysv init scripts.

The worst case scenario ( beside doing nothing ) is to have me to do both the migration and the packaging since I will have to stop migrating legacy sysv script to native systemd units and start playing catchup and start shipping already existing units ( I'm half way through the remaining alphabet of components [1]).

I should probably also mention the unit files I have created and submitted during the F17 reflect the current state of the project as in...

a) PIDFile= points to /run/foo.pid opposed to /var/run/foo.pid

b) binary paths should be reflecting the /usr proposal at this point in time

c) Environmentfile path that point to /etc/sysconfig/foo have been dropped ( or rather not included ) since those units wont fly with upstream due to not being cross distribution distributable which means in the long run if we are going to continue to support this each component would need to carry a patch against upstreamed unit that adds this behaviour to them and in essence this serves no purpose any more since administrators and experienced users should really be familiarizing themselves with [2].

1.https://fedoraproject.org/wiki/User:Johannbg/Features/SysVtoSystemd-F17

2.http://fedoraproject.org/wiki/Systemd#How_do_I_customize_a_unit_file.2F_add_a_custom_unit_file.3F

adding meeting keyword

I agree Viking_Ice aka Johann is working on unit files heavily and he did a lot of a good work. But I don't like Johann's interaction with other Fedora developers. Good example is here:
https://bugzilla.redhat.com/show_bug.cgi?id=699040#c39
I don't think pushing units instead of sysvinit scripts without closer interaction with developer would help.

-1 from me for Johann as provenpackager

Yes I did not like to dance around Steve's attitude either splitting those inits apart glueing them back together countless times over and over again and I lost my patience with him.

It was not a without a reason that I asked if another maintainer was available to deal with which is the first and the only case I have had to do that.

If he did what I pointed out to him to do in the first place bug 769879 which is afaik is still valid which means NFS IS BROKEN IN BOTH F17 AND RAWHIDE ( just to show for the record people can draw their own conclusion about his attitude against Fedora and it's user base).

In the future I suggest that you should really bother checking all the facts and weigh in all the bad and the good cases before reaching a decision.

If you want to turn this into some witch hunt don't bother with approving this request and just close this ticket I did not request it anyway nor was consulted before it was done...

This request was not approved at the 2012-02-27 meeting.

Login to comment on this ticket.

Metadata