On Sat, Sep 27, 2003 at 01:45:05AM -0700, Tom wrote: > On Sat, Sep 27, 2003 at 03:13:45AM -0500, Ron Johnson wrote: > > On Sat, 2003-09-27 at 02:54, [EMAIL PROTECTED] wrote: > > > On Fri, Sep 26, 2003 at 09:47:07PM +0200, Stefan Waidele jun. wrote: > > > > Howell Evans wrote: > > > > > > That's amazing; I would have expected cron to have information to know > > > a priori when to wake up, rather than "polling"... > > > > Waking up once a minute (which is a signal itself) to spend a 1/2 > > second reading some 5 line (yes, yes, they can be longer) text > > files is a *lot* simpler than having a signaling system for each > > crontab file. > > I figured that is the argument -- too much work for too little payoff -- > and that's okay! > > > > > > At the least it could optimize out obviously simple cases... > > > > Like what? > > > > Like the empty set. > > Think about it: if a guy tells you to call him every hour, and a guy > tells you to call him every fifteen minutes, do you think every minute > about who to call? >
Or, thinking a bit, why not alloc 1440 bits, and once a day process all the files, setting a bit for which minute to wake up, and when I wake up, walk forward for the next set bit, and sleep that many minutes? Do it all again if the files change... -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]
