Dear Axel Dörfler, Thanks for your comment and document. I'll more than happy to re-factoring the tty layer with following ideas - usable by other devices - not necessary compatible with BeOS How about architecture of this the new tty layer? Do we have a master design or I have to following some guidelines to make it work with others part software or current working device drivers. Regards, Ha ________________________________ From: Axel Dörfler <axeld@xxxxxxxxxxxxxxxx> To: haiku@xxxxxxxxxxxxx Sent: Wed, March 24, 2010 3:11:51 AM Subject: [haiku] Re: GSoC - Drivers on TTY Layer "François Revol" <revol@xxxxxxx> wrote: > Le Tue, 23 Mar 2010 14:20:22 +0100, Axel Dörfler a écrit : > > "François Revol" <revol@xxxxxxx> wrote: > > > Indeed, it shouldn't be too hard to take the current tty driver > > > to > > > make a module out of it, compatible with the BeOS tty module. > > It should definitely not be compatible to the BeOS TTY module. The > > API this one exports should be banned from this earth. It's even > > worse > > than Win32, and that says a lot. > Rico 0wnZ ! Oh well :-) > > > Though it > > > might be more interesting to convert the serial (pci and usb) > > > drivers > > > to the pnp stack, and write the tty layer as a consumer for the > > > nodes > > > published by those drivers, and producer for devfs. > > You are using a nomenclatura that Thomas original driver > > architecture > > was using, but that one has nothing to do with ours anymore. > Yeah, well, what's the official vocab for it btw ? First of all, it's just called device manager; the "pnp" has been discarded ages ago. And then, there is a device node tree with parent/child relations. Bye, Axel.