On Tue, Jul 16, 2013 at 12:39 PM, Miklos Szeredi <miklos@xxxxxxxxxx> wrote: > Is the border always where it should be? Are there any candidates for > moving things into, or out of the kernel? Are there some guidelines > for deciding which functionality belongs to the kernel? Now I need to flag ATTEND. On this subject I would add not just broadband modems but the media players and all type of Android devices that nowadays use the MTP protocol (and thus the PTP protocol). I am also the maintainer of libmtp. It has turned out that people think about this transactional object thing (which is more like FTP over USB than USB mass storage) as a file system and it is now being implemented with e.g. GVFS and KDE's counter-part in userspace as virtual filesystems. This would be all OK if it wasn't that I always have this worry that with all the device-specific quirks etc we had been better off in kernelspace for what is today libmtp. Certainly we had gotten to a point where it was usable for people quicker than has happened so far. The strangeness of how libusb abstracts things to userspace would not be there, and it would be more possible to share code between client and server (responder and initiator as they are called) than is currently possible. So I guess I could add some €0.01 here... Background: http://libmtp.git.sourceforge.net/git/gitweb.cgi?p=libmtp/libmtp;a=blob;f=README;hb=HEAD (The end of the file is especially interesting, may propose this as a talk for a conference this year.) Yours, Linus Walleij -- To unsubscribe from this list: send the line "unsubscribe linux-usb" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html
