OK, most of you are probably sick and tired of this, but there have again been some pretty major changes to bttv2. If you feel like giving a hand, please get it from the CVS (http://bttv-v4l2.sourceforge.net/cvs.html), and let me know if it works (or more likely, what doesn't work). What's new: 1) Dynamic RISCMEM length. Most risc tables can fit in a single page, and single pages can also be allocated easier, so this should help a log if you are using a lot of capture buffers. 2) THE BIG CHANGE... Integrated multiple capturing opens. This needs some serious testing, and also some clarrification of the API. How it is implemented: * The device context contains the info on inputs, standards, frequencies, etc. * The device context also contains a single preview resource (can anybody see a need for multiple previewing opens???). * Each open has its own capturing resource, which can be configured independantly of each other. * At STREAMON time the resources required for the capture are locked down (or if they are already in use -EBUSY is returned). * At STREAMOFF time the resources are freed again. I think this is more or less how the API discussion ended, although matters were getting a bit confused towards the end... Any comments/suggestions/flames^W/test results appreciated. -justin
