Jack Pryne wrote: > [cut] > > Imagine a P2P network of Debian users who *all* shared their system > configuration information, thus comprising a consensus of proper > configuration for various hardware compositions, (e.g. All users with the > AsusV7 motherboard and Matrox G400 graphics card would share information to > devise the best configuration for that hardware.) > > Once a user-base is established from existing Debian users, new users could > download this small P2P application. Upon running it for the first time, it > would set about making a partition used expressly for Debian, and start > downloading the Debian operating system. Once the basics were in place, > hardware self-diagnostics would begin, and searches across the P2P network > would reveal other users with similar hardware setups. Then, actual > configuration would take place automatically. This new user would recieve > 'advice' (from other users with identical or similar hardware) about which > drivers to download and where to place them. Properly planned, such an > application might well automate the installation and maintainence process! > There's some merit in this. Maybe you could eliminate some of the more difficult and scary parts, like self-diagnostics, creating the partition on the fly, etc., at least at first. I.e., do what you can with the infrastructure that already exists.
What I'm thinking: (1) Set up a project (SourceForge?) whose goal is to create .debs for many common peripherals. The debs might include kernel modules, and things like modeline settings for monitors, magicfilters for printers, on and on. You really need some Real Debian Developers involved here. Note that these debs would have *many* dependencies on debs in the main Debian distribution. Hmmm, I guess they really should be task packages. Like task-epson740. Some thought needs to go into what extra prebuilt kernels, if any, would be needed. (Think about the excellent linuxprinting.org, and ask "What if it was Debian-oriented" and "What if it pulled in all kinds of peripherals?"). Of course, this part could also be done by Real Debian Developers as a part of the Debian project. Or one Real Debian Developer sponsoring a lot of other folks. They might not want their mirrors clogged up with all of this, though. (2) The little Windows app might be one to scan the registry and see what all Windows recognizes as being on the system, and write that to a file. (I mean, presumably you've got all your stuff working on Windows already, right?) Need a Real Windows Developer here. (3) Have a cgi app on your website that uploads the file and uses some searching voodoo to find a best match for debs. Need a Real CGI/{Perl|Python|C} Developer here. The user might need to do some reasoning/research to figure out which selected debs actually are appropriate. (4) The user installs a basic Debian system, then uses the selected debs to do the correct remaining voodoo for his peripherals. Steps 1 is the really large task requiring huge Debian experience and expertise. Whether it's usefully doable at all is an open question. 2 should be really easy; 3 might not be too bad. Enough handwaving for you? :-)