okay:

- different CPU time % load aka throttle is ineffective. Only works to reduce temperature and does nothing for system responsiveness.
- % of RAM Memory during use and idle can already be set.
- Crunching during idle and use can already be set with a delay factor

What would be a good feature, multiple requests preceding, is to have a % CPU Time and % CPU cores during use and idle (fan noise for instance when at the desktop).

I'm not up to speed as to what's in the development pipeline at Berkeley, where client features are discussed and implemented. Here I'm [others too] thinking e.g. for a gamers quad machine to have 2 cores reserved for that and the other 2 cores continuing on crunching. There is already a function to stop BOINC automatically when applications of choice are loaded in memory.

Automated work fetch scheduling in consideration of a vacation/trip schedule... definitely not easy and needs proposal at Berkeley... including the basic logic to include multi-project considerations. Presently practically, with all WCG production deadlines of 10 days and knowing you have a trip, flip the connect and or additional buffer setting. Set client to not fetch work, purge anything not complete just before departure and allow work fetch again upon return. For shorter than deadline trips, certainly set buffer to a size that is deadline minus trip duration so as to be able to complete /deliver results upon return, whatever remained incomplete. But, if you go on a trip, how much of the daily time would be available to crunch off-line? Only you would know and it still being a guess.

But, since you're IBM and WCG is run by IBM, maybe the techs see something that could be added to the web profiles as coding this and getting it implemented in the client needs an divine act... and it has to be done in such a way that it does not upset the old clients that don't have the features... I see the word trouble painted in that limitless sky :D