On Fri, 5 May 2000, Ron Stodden wrote:
> Guillaume Cottenceau wrote:
> >
> > "Hoyt" <[EMAIL PROTECTED]> writes:
>
> > > > Can you imagine the reaction of the people if we make a "updatedb" at the
> > > > end of install, blocking the system for 2-3 minutes.. When you wait and
> > > > nothing happens, 150 seconds are really long seconds!
> > >
> > > Here is my reaction: I am investing a significant amount of time in the
> > > install - another 3 minutes to run these commands is not too much,
> > > especially if it would enable "a half-dozen useful/essential commands". All
> > > you have to do is let me know what is occurring via a messsage on the
> > > screen.
> >
> > Don't forget we are talking about "updatedb", not "rehash"!
>
> We are talking about running updatedb and makewhatis.
>
> > Sure. I still think that the best solution would be to launch "updatedb"
> > as background task at first boot ; not to *block* the installer for
> > something you certainly not use during the first seconds of your first
> > login.
>
> Cannot agree. 3 minutes at first boot is much more glaringly obvious
> - and disconcerting (the rule is: No surprises!) - than 3 minutes
> added to the install time.
I agree with you on that - it does look bad.
But why not let the user decide?
Maybe have an extra stage asking:
Would you like to initilize system databases [pick a less bombastic
description] now (this will take a few minutes)?
and the help text will show how to manually initilize those manually.
Another possible solution is to make the first boot a part f the install
procedure. But I guess that it would be problematic to make it both
user-friendly and general enough.
>
> There is another unreporeted problem here:
>
> updatedb is scheduled by cron at 0402hrs, so it is only done for
> machines that arerunning at that time, which for personal computers
> (ie Linux) is hardly any. It should be invoked through anacron, not
> cron, so that it is always run every new day that Linux is invoked.
>
> Will this problem be accepted and logged (and a confirmationnumber
> sent to me) for fixing?
But wouldn't such an invocation also be considered a surprise?
--
Tzafrir Cohen
http://www.technion.ac.il/~tzafrir
