[Date Prev][Date Next] [Thread Prev][Thread Next] [Date Index] [Thread Index]

Re: Fwd: Re: cron in UTC?




2014/10/01 21:29 "Henrique de Moraes Holschuh" <hmh@debian.org>:
>
> On Wed, 01 Oct 2014, Joel Rees wrote:
> > Should I use this as my excuse to actually join the dev team, in spite of
> > my misgivings about systemd and the API creep?
>
> Only if you promisse me you are never going to mention systemd again on the
> communication threads where fcron work is taking place, except for the bare
> minimum actually related to fcron. 

Heh. No, that wouldn't really be a problem.

> It will need a service file and an
> initscript because it has to be started on boot by both systemd and
> sysvinit.
>
> What is a lot more troublesome is that someone in the team will need to test
> it SELinux mode.  And if Debian ever adds AppArmor, fcron will have to
> interface to it as well.  cron-like daemons are security-sensitive packages.
>
> > Is your third going to need to run jessie with systemd? How much and what
> > kind of hardware/OS resources does he or she need to be able to bring to
> > the table?
>
> You will need an unstable chroot, and you must test the packages there, as
> that's the maint target for integration.  This doesn't mean your main system
> must run unstable.  IME chroots and VMs are enough to deal with this.
>
> And yes, it has to be tested with systemd and sysvinit, in both cases with
> SELinux enabled.

There's the real problem, and the one that has stopped me in the past -- hardware to set up the dev and test environments. The hardware I have might be able to handle chroots, but it won't do VMs. Too old.

Guess I was thinking maybe there would be something I could do in an alternate boot partition on an older 32 bit machine. I have to do something about my hardware, but I don't have the extra money at this point.

> So, it is a reasonable amount of work to do it properly.  THIS is why I am
> very upfront about the fact that it needs several maintainers with time to
> spend on it, and that I won't be able to do much other than coordinate right
> now.
>
> We can very reasonably expect the time sink to get a lot better after the
> package shapes up, as fcron development is slow (and it looks like it has
> picked up some, so it is out of maintenance-only mode!).  It is also pretty
> clear we have to track upstream development closely, and cherry-pick patches
> from the ML.
>
> However, I do recall that maintaining fcron packages was a lot of fun. 

Yeah.

> You
> *really* grow as a maintainer when you take care of something like that.
> And fcron is a real cool piece of software, although I expect that other
> rather nice vixie-cron replacements must have matured in the last 10 years,
> so it would also make sense to check the competition first, before spending
> a lot of maintainer resources to reintroduce fcron in Debian.

Well, maybe some one on the user list with more free resources than I could get  up for some fun.


Joel Rees


Reply to: