[sugar] human factors when launching an activity
Wed Oct 1 15:06:34 EDT 2008
Martin, your response has brought up a number of topics:
> About two weeks ago I had this same impression as you do now, though,
> before I noticed that it was because the datastore was busy indexing a
> few kernel build trees on two usb drives I had plugged in.
Well, I have a NUMBER of removable storage devices plugged in. When
I first started, I wrote a ticket saying "please give me a way to
tell Journal NOT to 'index' these". That ticket was closed with:
"that's something we have in mind <for a future implementation>".
As it is, when I do an 'olpc-dump', I've seen it be larger than 1MB
-- because of the 'datastore.log' files. Obviously, *something* is
spending a LOT of time "walking" my removable storage devices. But
since this is something NOT DOCUMENTED for users to control, I'm
choosing to live with it. [And complain about Sugar responsiveness.]
> Are you sure that the machine is otherwise idle when you're
> experiencing these undesireable delays?
How should I know ? Is there a 'resource monitor' for Sugar ?
If I walk up to the machine at a random moment and issue 'top', it
typically shows the machine to be 90% idle. But there have been
plenty of times when 'top' shows it to be 0% idle. For instance, if
I do 'sugar-control-panel -g available_updates' on a recent Joyride,
the machine goes 0% idle for 10+ minutes, then gets an OOM.
> dbus-send --session --print-reply --reply-timeout=3D2000 \
> --type=3Dmethod_call --dest=3Dorg.laptop.sugar.DataStore \
> /org/laptop/sugar/DataStore org.laptop.sugar.DataStore.unmount \
> string:<datastore mount id>
THIS is what I have the greatest amount of difficulty with. I fail
to see the necessity of users having to learn the "internals" of
dbus usage in order to be able to control their systems. If there
is a "knob" that users need to turn, DOCUMENT IT !!!
Besides which, I have my removable storage devices mounted __for a
purpose__. [They contain software and data which I reference every
time I use the system.] I have no intention of unmounting them.
More information about the Sugar-devel