I think this would be a great option, as I use chat programs for work.
+1
On Jul 9, 2010 11:48 AM, "Stefano Z." <ditta...@gmail.com> wrote:
+1
On Fri, Jul 9, 2010 at 8:06 PM, D.r.a.g.o.n <d.r.a.g...@gmail.com> wrote:
> I think this would...
I utterly disregarded the wise advice of focusing on fixing bugs
before attempting crazy stuff like this - but for a good reason!
(which is, I was - and still am - going insane with the whole UJ
situation and needed a break; and what's more relaxing than attempting
to write improbable-looking code in a completely unknown area?)
So, find your beloved and repeatedly +1ed 'Apps' trigger in v1.8.2! -
and, if it works, in later builds as well (fingers crossed!)
Technicalities: there's indeed no way of telling a 'useful' background
app from a 'useless' one - one which is there just because it was
"cached" by Android or because it went ahead and started its own
service which just sits there with the sole purpose of wasting RAM (as
99.99% of apps do).
Soooo, the trigger *only* takes into account *visible* apps - apps
that are either currently in the foreground (i.e. you're actively
interacting with them), *or* that registered a notification in the
status bar. Hopefully the latter category will take care of all the
cases where it makes sense to keep the connectivity enabled (e.g.
music streaming apps with their 'play' notification, or ongoing chats,
or the like) - while the former one will handle the 'blacklist' case
(games, offline reading, etc.) nicely.
As a quick aside, the CPU trigger might now work in unintended ways
(most notably underclocking the CPU while you play 'blacklisted' games
- i.e. at the worst possible time) - so it'll need some rewriting.
It might be a good idea to keep the discussion for the new trigger on
this thread, as newer builds will have to deal with other fixes and
improvements (most notably Location).
On Fri, Jul 9, 2010 at 11:42 PM, Nitin Philip <nitin...@gmail.com> wrote:
> That's true, JD could, in theory filter out the 'cached' processes and only
> take active (visible, secondary service etc) processes into account. I have
> no idea if this will actually work or cause more problems, hopefully Mark
> can work his magic and figure out something. (I really need JD to stop
> turning my WiFi off when Pandora is playing)
> I also agree that the priority right now should be getting JD to work
> reliably on the bazillion Android phones out now, and future CDMA based
> phones too..
>
Sounds great :)
Enviado desde mi "peazo móvil"
> That's true, JD cou...