> * I'm not sure whether it is a project big enough for GSoC. The file system > itself will be relatively simple, so I think the basic framework can be > implemented in a few days; maybe add one or two more to get familiar with > Haiku's FS interface. The is the main task would have to be the > implementation of the different services offered. ATM I don't have an > overview of all potential ones, so I can't really say much about that. > Maybe you can detail on which ones you're thinking of. Woo, I don't think it's just piece of cake to put a /proc filesytem here. You know most of the contenet of /proc files are generated on the fly. Maybe lots of jobs must be done for each sepecific proc file. > * In Haiku all file systems support node monitoring (i.e. anyone can > request to be notified when something in the FS changes). Since a procfs > shows statistical information that change over time, several kernel > services would have to notify procfs to provide that feature, and > potentially a lot of node monitoring messages would be sent. Performance > might be an issue here. I'm not familiar with the node monitioring mechanism. Is it possible to make the /proc filesystem outside of the node monitoring? > * The directory layout of /proc in Linux looks a bit ugly. Directories for > processes are mixed with all kind of other stuff. If we had to copy that > layout for compatibility, I'd be less inclined to like the idea. The layout maybe different, of cuz. :)