init observations

[Mike A. Harris]

Jeff Spaleta wrote:
> On 11/15/05, Dimi Paun <dimi at lattica.com> wrote:
> 
>>Basically, can we have X up in 10s? :)
> 
> 
> I still don't understand exactly why this is the important goal.
> Explain to me why the 10second to X goal on "reboot" is more important
> than getting a robust suspend/hibernate working that doesn't require a
> full boot up process at all? I can see why the fist 4 items in Luke's
> list are technical wins for a less grotesque init process... but i
> still don't get why getting to a login screen in under 10 seconds on a
> full boot is noteworthy or highly desirable compared to a
> suspend/hibernate that actually works across desktops and laptop
> hardware.
> 
> And until we see bootchart output that shows a comparison of initrd to
> SysVinit for the default Fedora Core service set... the jury is out on
> how big a difference this actually makes for the default case.

"X" starts in less than 5 seconds on most systems and has for years.
However, if what is meant is "Have a fully working graphical GNOME or
KDE desktop started up in X in less than 10 seconds" is the goal, then
my opinion is thus:

Getting a full desktop started up in less than 10 seconds is a
completely orthagonal problem than having working suspend/hibernate
support.  Most likely the people who would be working on either
project, would be different developers working in parallel on
different problems.  So it isn't IMHO about allocation of resources
per se.

I personally think that getting a full desktop running in less than
10 seconds is more important than working hibernate support, simply
because there are probably one ore more orders of magnitude more
"desktop users" than there are "laptop users" out there.  Both
features are important to have for their respective userbases of
course, but the "desktop users" group generally is a superset of the
"laptop users" group.