Idle-suspend a little too intrusive to user experience?

Paul Fox pgf at
Mon May 10 10:38:25 EDT 2010

daniel -- problems noted.  i agree with the team that increasing
the timeout to a minutes seems prudent for the time being.  i have
some ideas on how to make inhibiting suspend more robust, and
i've been working on the dcon issue(s) (#9631, #9664) in the last
day or so.

daniel wrote:
 > Laptops have not yet been handed out here in La Rioja, but a few have
 > been handed around to various parts of the MinEd and support staff, so
 > we're already seeing some bug reports and user experiences rolling in.
 > The idle-suspend experience has been causing some discomfort.
 > Specifically 4 cases:
 > 1. The machine suspended while playing a video in Jukebox (probably
 > not a common case, but it happened, and I had to explain this system,
 > and it's not something that's very easy to explain to non-techies!)

i'm quite interested in this one, because it's one i really thought
shouldn't happen.

 > 2. The machine suspends frequently while it is loading a web page, and
 > does not wake up, meaning that the web page doesn't finish loading
 > until you realise what's happened and intervene

i'm going to venture a guess that the network there isn't quite
as responsive as it is here at 1cc.  right now we inhibit based
mostly on outbound network traffic within 5 seconds of the
suspend point.  in the case of slow DNS or web-server response, i
can imagine that that might not be sufficient.  the mechanics
within powerd for doing the detection are a little cumbersome
right now -- i think i know how to fix it, but it's non-trivial. 
if i can rework it, it will be much easier (i hope) to extend the
detection interval.

 > 3. After a lid-suspend, the machine frequently sleeps before it has
 > scanned or reconnected for wireless networks. The user sits there
 > waiting and waiting for the network to appear again, and it's not
 > going to appear until they realise what's happened and intervene.
 > (related #9854, but I feel that this is a more general problem)

i'll try and find a way to detect that we're in the "associating"
phase.  this one has hit me, too, and i agree it's very annoying
to realize you're waiting for something that's never going to happen.


 > 4. Sometimes a stale version of the screen is loaded as the machine
 > goes into suspend - generally the screen that was locked upon the
 > previous suspend. This is really confusing for the user. (#9664)
 > As a result of these early issues, the team here wanted to increase
 > the idle suspend timeout from 15 seconds to 15 minutes. I managed to
 > persuade them not to do this until we've actually seen how it holds up
 > in the hands of actual users (children) but in the mean time they will
 > be increasing that timeout to 60 seconds in order to soften the blow.
 > Daniel
 > _______________________________________________
 > Devel mailing list
 > Devel at

 paul fox, pgf at

More information about the Devel mailing list