On Tue, Oct 09, 2001 at 10:20:11AM -0700, Jean Tourrilhes wrote: > I just have a few comments... > 1) There is more USB-IrDA devices than those you list in the > driver, and most IrDA-USB devices don't follow the spec in some way or > another. Please check .../drivers/net/irda/irda-usb.c for the gory > details...
I know. I only had access to that one device. I'll be patching the driver for the other devices soon. > 2) The idea is great, but I think the overall architecture > could be improved. The way you have done it leads to code duplication > (irda-usb.c + ir-usb.c) and work only for a small subset of the IrDA > devices (i.e. don't for for other FIR devices such as NSC, SMC, ALI, > SigmaTel-USB...). The API of IrDA driver is simple and well defined, > so it would be IMHO much more efficient to do a simple TTY wrapper > that would work over *any* IrDA driver. I think that LIRC is moving > this way... The only duplication is a bit in the logic to read the irda config. I only want to provide support for any USB IrDA devices. A tty driver wrapper for all IrDA devices would be great, but I didn't see how to easily do that (remember, no new major/minor numbers are being allocated.) So for now a simple usb-serial driver for these devices meets a need. > 3) The Linux-IrDA stack already allow you to talk to embedded > devices with no IrDA support. This is called the Ultra protocol (very > simple, no handshake, no management frames). I can give you some > examples if you want... Pointers to this would be great, thanks. > 4) You forgot to use USB_ZERO_PACKET. A bit of testing would > have shown you that this is necessary. Why do you think I spent so > much time bothering this list about it ;-) Ah, I don't see how I missed that one. Thanks a lot for the pointer. My testing was very minimal. greg k-h _______________________________________________ [EMAIL PROTECTED] To unsubscribe, use the last form field at: https://lists.sourceforge.net/lists/listinfo/linux-usb-devel
