Christian Packmann wrote: >> Actually, the kernel now uses node monitoring to detect driver >> updates; "rescan" should be a no op. > Oh. The transcript of your presentation at WalterCon 2006 indicated that > rescan was supposed to be used. That part hadn't been implemented in 2006 yet :-) >> BTW the code for the legacy driver support is in src/system/kernel/ >> device_manager/legacy_drivers.* > Uhm, what is the distinction for "legacy driver"? I modeled my new > driver after poke because I thought it would use any new model, being in > the Haiku repo. Is there any driver which is a.) simple and b.) a good > representation of "how it shall be done"? Legacy drivers are using the BeOS driver model. This is what most drivers are still using; and since the new driver architecture isn't really finalized yet, there is nothing wrong in using that one for now. Bye, Axel.