I probably have to add signature - I do not receive Cooker. Cc to me if you want me to respond.
> just a short and dumb question, > how will be chosen the configuration method to use(2.4 or 2.6) I do not understand the question. configuration method is drakmouse or whatever. it does not depend on kernel. but it must emit something suitable for both kernels. > with oder words, what kind of problem should expect in case > i'm running 2.4 with the input from 2.6 > in a not that common way: local XFree multi-user you can't. if you running 2.4 you are running input from 2.4 > PS. > about PS2 devices: are they going to be loaded even if not present > (you mentioned modprobe.conf) I mentioned modprobe.preload. Credits to maintainers who still not updated mkinitrd and initscripts. they are going to be loaded if they are configured to be loaded. So far we do not have hotplugging for these devices so they are going to be configured statically yes. > about mice alocation: we have a very hackish input.agent which works > but ... is really hackish and probably wont work both with std 2.4 > and 2.6 kernel. > http://varna.demon.co.uk/~svetlio/aivils/mdk/input.agent good, I'll get a look. I probably integrate it into devfsd though. hotplug is not the right place to do it in presence of devfs. input agent must load input handlers and it will do it this week with depmod help. your input agent actually does the same as udev. I am seriously considering snathing namedev out of it as devfsd action handler. anyway we have problem - user most likely installs and configures mouse under 2.4 so 2.6 specific information (like DEVPATH) simply is not available. [...] the rest omitted due to inability to do proper quouting using copy'n'paste from web archive. -andrey
