[08:49:57]  <kay> Keybuk: m-i-t: you still playing with the udevd-less
   ideas? or what's the reason for these rather "weird" questions? :)
   [10:39:58]  <Keybuk> kay: nah, not really relevant to udevd-less ideas
   [10:40:12]  <Keybuk> just trying to understand the use case
   [10:40:14]  <Keybuk> I think I've got it now
   [10:40:25]  <Keybuk> it's for when you don't build ppp into the kernel,
   you need something to make /dev/ppp
   [10:40:31]  <Keybuk> it doesn't cause ppp to be loaded though as a module
   [10:40:36]  <Keybuk> it just makes /dev/ppp - right?
   [10:40:43]  <Keybuk> so when you open that, the kernel still does the
   char-major thing?
   [10:41:25]  <kay> yeah, exactly
   [10:41:33]  <singesang> hello
   [10:41:40]  <kay> it replaces the /lib/udev/devices stuff in many cases
   [10:41:41]  <davidz> Keybuk: kay is just trying to make udev work like
   devfs did ;-)
   [10:41:53]  <kay> sure, that's the plan :)
   [10:42:02]  <kay> and it works fine, as it seems :)
   [10:42:21]  <kay> Keybuk: so the kernel can supply static nodes for
   on-demand loading
   [10:42:36]  <kay> no need for a fuse init script and such
   [10:42:47]  <kay> /dev/fuse will trigger the load of the module
   [10:43:09]  <kay> udev will mount the fuse control fs when the event for
   the misc device comes it
   [10:43:12]  <kay> comes in
   [10:43:31]  <kay> same for device mapper, btrfs, tun, ...
   [10:43:53]  <davidz^ oh noz you are taking away customizability and trying
   to make all distros look the same! like.. things are not supposed to
   _just_ work! It's supposed to be HARD!
   [10:43:54]  <kay> all will just magically activated when the module is
   there, but not loaded
   [10:44:21]  <kay> davidz: oh we will find really hard stuff to throw at
   people :)
   [10:44:46]  <davidz> no, we're supposed to be solving the same problems
   over and over again ;-)
   [10:45:02]  <kay> we still do, don't we? :)
   [10:45:27]  <kay> davidz: we will get in-kernel media presence polling :)
   [10:45:37]  <davidz^ sounds good
   [10:45:58]  <davidz> yes, it's all one eternal rewrite
   [10:46:07]  <davidz> moreso now than ever...
   [10:46:11]  <kay^ hehe, yeah :)
   [10:46:27]  <kay> but as long as the new stuff looks better to us :)
   [10:46:46]  <kay> i'm pretty happy with what we did so far ...
   [10:46:53]  <kay> it seems to work fine here
   [10:47:06]  <kay> i even did "rpm -e hal" just fine this week
   [10:47:22]  <Keybuk^ yeah, I still thought you loaded the module as well I
   think - thus trying to put that in module-init-tools
   [10:47:51]  <Keybuk> davidz: oh, that's ok - with the different init
   systems, no two distros will ever look the same again <g>
   [10:47:53]  <kay^ and the nice thing is that if ther e is no module, or
   it's compiled-in, there is no static node
   [10:48:03]  <davidz^ doesn't yet work here... Remove       30 Package(s)
   [10:48:03]  <davidz> Reinstall     0 Package(s)
   [10:48:04]  <davidz> Downgrade     0 Package(s)
   [10:48:04]  <Keybuk> kay: we've worked around all those so far by just
   building them in <g>
   [10:48:22]  <Keybuk> yeah, now I get it, it's kinda sweet
   [10:48:27]  <kay^ ah, i see. yeah, the trick is to laet the stuff only be
   loaded on-demand
   [10:48:51]  <davidz^ well, it does but I don't want gnome-panel or
   gnome-session removed... ;-)
   [10:48:54]  <Keybuk> the trouble is though you still end up with a period
   after the kernel is loaded, and before starting udevd, where /dev might
   not be "ready"
   [10:48:57]  <kay^ it even applied the permissions from the rules files :)
   [10:49:38]  <kay> yeah, but before udevd is started you probably don't
   need fuse, tun, ppp, ... :)
   [10:49:53]  <kay> it does not depend on trigger, just udevd
   [10:50:02]  <Keybuk> right
   [10:50:06]  <Keybuk> kudev!
   [10:50:15]  <kay> davidz: really? that wants hal?
   [10:50:33]  <Keybuk^ one of the bits still "needs" HAL for brightness
   control of displays
   [10:50:37]  <Keybuk> it might be gnome-hughsie-manager
   [10:50:54]  <singesang> i've got a umts stick which needs a usb mode
   switching. that works fine for me. kernel's usb<->serial driver also
   creates the corresponding ttyUSB[0-3] devices and two symlinks from
   /dev/gsmmodem -> /dev/ttyUSB3 and from /dev/gsmmodem3 to /dev/ttyUSB0. but
   last symlink doesn't always gets created. how can i make it, that always
   the right devices get created?
   [10:51:03]  <kay> Keybuk: ah, we removed that dep, it will just fail if
   hal isn't installed
   [10:51:19]  <hughsie+ g-p-m in git master provides a fallback to sys
   [10:51:23]  <kay> singesang: look in /dev/serial/?
   [10:54:13]  <Keybuk^ we made HAL service bus activated
   [10:54:48]  <singesang> erm yeah. stupid me. i forgot that. thanks for the
   hint :)
   [10:54:55]  <kay> Keybuk: yeah, i know. makes sense. but i wanted to get
   rid of it entirely, and it worked :)
   [10:55:45]  <kay> stuff like non xarandr can still install it. but it
   should at least de-installable :)
   [11:03:04]  <Keybuk> *nods*