On Tue, Apr 08, 2008 at 07:33:41AM +0200, Benoit Donnette wrote: > First of all, thank you for your amazing work, Greg. > > Now one more element : > > Quite a few USB/IEEE1394 devices are in fact supported by third party > software usually not coded as kernel drivers, and a lot of them do not > bother not being Linux-aware. The one I am most interested in is digital > cameras (DSLR, to be precise), for all the things not related to image > transfers (these use the Mass Storage protocol most of the time). Using a > DSLR controlled from a computer requires a Win machine or a Macintosh > today, and this is heavily used in studios. This is the last obstacle for > full-Linux image companies. We could then suggest a few breakthroughs > (revision control integrated in a document management system for example), > and these could make the difference. > > I'd have some time to spend on it as soon as I get a camera. I'll try to > contact the Open-Source friendliest manufacturer first (Pentax, whose lens > mount is mostly open, who allows for a documented raw format : these are > fair signs of openness). We have the ability to control firewire devices from userspace through the "raw" driver, right? And for USB, we have libusb and openusb, both cross-platform userspace libraries built on top of usbfs. So from the kernel point-of-view, I think that there's not much we can do here, right? I've pushed many USB devices to userspace programs using libusb as that is the proper place for them (they don't need to be kernel drivers). This gets the benifit of working on all OSes that libusb supports (solaris, *bsd, windows, os-x, etc.) If you know of specific devices that are not working on Linux, _and_ whose manufacturers are interested in changing this, please let me know, I'll be glad to help out. thanks, greg k-h
