Re: Announcing bcron version 0.08

On Mon, Jun 13, 2005 at 03:26:59PM -0600, Bruce Guenter wrote:
> On Sun, Apr 03, 2005 at 07:24:03PM -0600, Bruce Guenter wrote:
> > > Hmm, the bcron-update service still bugs me somehow.  It could be
> > > replaced with a magic file in /var/spool/cron/crontabs/ for example, so
> > > saving one service, having the system crontabs updated through a program
> > > run by bcron-exec.
> > You know, that even makes sense.  I'm not sure why I didn't do it that
> > way myself.  bcron-update already updates the sched through the trigger,
> > so changes are scheduled in the next minute after the update.
> 
> One problem with this is that, in order to determine what files are new,
> bcron-update will have to open and read all the directories involved,
> which will cause atime updates every minute.  This is a pain for systems
> such as laptops, which will want to spin down their hard drives.  I can,

Yes, I see.  So that's why you originally did it this way ;-).  Now I
think it's good to have bcron-update a separate service, but even more
dislike the cron standard's requirement to check for these files every
minute.  Better would be to commit changes to the system crontabs
through another program, just like newaliases for /etc/aliases, make for
tinydns' data, or init q for /etc/inittab.

> however, add a "one-shot" flag to -update so that it can be used both
> ways if this is still desirable.

That would be nice, better yet IMHO would be to have a flag to run it as
service, and have the "one-shot" by default.  One then can remove the
service, and always simply call bcron-update after changing the system
crontabs.

Regards, Gerrit.