Screw NetworkManager, its broken anyways and wpa_supplicant can already do everything one might want there:
Add 'mac_addr=1' and 'preassoc_mac_addr=1' to your/etc/wpa_supplicant.conf. Then your MAC-address will be randomized during the Scanning/Preassociation phase and afterwards.
For networks that need a static MAC address for filtering, add 'mac_addr=0' in the appropriate 'network' section. You also want to make sure you are using 'dhcpcd' instead of 'dhclient' (alias isc-dhcp-client). The latter can't deal with changing MAC addresses, it seems.
I don't think getting rid of what you mentioned as 'bloat' will really buy battery time. I suggest you get powertop and see where all those wakeups are coming from. Nice citation from there: "A good thing to try is killall gnome-power-manager". How paradox. If you follow the tips on their site, 20% more battery life are easily achievable (3.5h instead of about 3h on my T60).
The main problem with power consumption is a "We no longer care about CPU cycles" attitude among many programmers, especially among the KDE and Gnome crowd. Why is there a daemon for every little thing programs could formerly handle by themselves or through libraries?
Like gconfd for parsing configs and watching them for changes. Or dbus, as if there were no othere proven methods for IPC, that don't require another daemon idling around and waking up every other millisecond eating away battery life. Or just log out from a KDE session and watch those 10 or so beauties like dcopserver idling aroud, eating memory. And does anybody even know what something like bonobo-activation-daemon does?
The laziness of application programmers has gone much to far, instead of using methods that are provided by the operating system and just require finding them there is a load of new, redundant mechanisms mostly implemented by new daemons. Every programmer introducing some new battery-eater should be required to justify this additional power by more than just "its easier this way", "windows also has some registry-parsing daemon" or "but I don't like parsing sysfs myself".
NO_HZ is nice, but only curing the symptoms of a larger problem: daemon-bloat! Get rid of them and you will see some real improvements.
...because MS might do something? Right!
MS is proven to have a monopoly on at least one market
(Desktop-OSes, Office-Software, etc.). So in Europe
the comission has to check every major action that company
undertakes, like mergers or entering new markets. They
might even decide to limit microsoft's actions up to a
certain degree or even deny them at all.
This might seem somewhat strange from an american point
of view. There it seems more like letting them do the mess
and clean up afterwards.
Slashdot Mirror
Screw NetworkManager, its broken anyways and wpa_supplicant can already do everything one might want there:
Add 'mac_addr=1' and 'preassoc_mac_addr=1' to your /etc/wpa_supplicant.conf. Then your MAC-address will be randomized during the Scanning/Preassociation phase and afterwards.
For networks that need a static MAC address for filtering, add 'mac_addr=0' in the appropriate 'network' section. You also want to make sure you are using 'dhcpcd' instead of 'dhclient' (alias isc-dhcp-client). The latter can't deal with changing MAC addresses, it seems.
The main problem with power consumption is a "We no longer care about CPU cycles" attitude among many programmers, especially among the KDE and Gnome crowd. Why is there a daemon for every little thing programs could formerly handle by themselves or through libraries?
Like gconfd for parsing configs and watching them for changes. Or dbus, as if there were no othere proven methods for IPC, that don't require another daemon idling around and waking up every other millisecond eating away battery life. Or just log out from a KDE session and watch those 10 or so beauties like dcopserver idling aroud, eating memory. And does anybody even know what something like bonobo-activation-daemon does?
The laziness of application programmers has gone much to far, instead of using methods that are provided by the operating system and just require finding them there is a load of new, redundant mechanisms mostly implemented by new daemons. Every programmer introducing some new battery-eater should be required to justify this additional power by more than just "its easier this way", "windows also has some registry-parsing daemon" or "but I don't like parsing sysfs myself".
NO_HZ is nice, but only curing the symptoms of a larger problem: daemon-bloat! Get rid of them and you will see some real improvements.
Hm, I would say, don't buy their stuff.
Err, but..., well..., ähm.
What do we learn from this? Standards need to be free.
...because MS might do something? Right!
MS is proven to have a monopoly on at least one market (Desktop-OSes, Office-Software, etc.). So in Europe the comission has to check every major action that company undertakes, like mergers or entering new markets. They might even decide to limit microsoft's actions up to a certain degree or even deny them at all.
This might seem somewhat strange from an american point of view. There it seems more like letting them do the mess and clean up afterwards.
Intel has a great compiler which is a lot faster than GCC or VC++ on x86 and ia64. It's even faster on my Athlon ;)
From: chief@the.apache.tribe.us
To: white.people@somewhere.us
Subject: This means war!
White men, either you give us Whiskey, guns or a wireless lan.
If not, you will die.