Grub and Plymouth

[Adam Jackson]

On Tue, 2008-10-14 at 19:52 +0000, Martin Sourada wrote:
> On Tue, 2008-10-14 at 09:37 -0400, Adam Jackson wrote:
> > That's the text plugin, yes.  It's not going to progress "smoothly",
> > it's ascii art.  We were hoping KMS support would land for Intel chips
> > for F10 but it looks like that's not going to happen.
> > 
> On a similar note, I noticed the progress bar finishes long before the
> boot process (I think somewhere around the time services start to
> start). This isn't an expected behaviour from user point of view.
> Actually it is pretty weird... I've installed readahead (not sure though
> if it is enabled, or how to enable it if not) and disabled some unneeded
> services which resulted in the progress bar showing up incomplete for
> about only a third of the boot process (and completing really fast).
> When I look at the real progress after the progress bar is already
> finished, I see services starting.

Since it is, in general, impossible to know how long service startup
will take, we lie.  We record the time it took to boot last time and run
the progress bar against that.  There's a detail here which is that you
can't know that until you've mounted /, so we estimate something really
long initially and then once we load the time off the disk we lerp
between the estimate and the recorded time.

We also cheat in that the growth of the progress bar is intentionally
nonlinear, growing faster early and then slowing down towards the end.
This fools your brain into thinking it's going faster than it really is.
Usually the math works out such that you'll see the bar fully complete
for a second or two before switching to gdm or text console.  But,
sometimes much longer, if this boot happens to take ten seconds longer
than last one then there's just not much we can do about that.

- ajax
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/pgp-signature
Size: 197 bytes
Desc: This is a digitally signed message part
Url : http://lists.fedoraproject.org/pipermail/test/attachments/20081016/76f1e391/attachment.bin