On Tuesday 13 March 2001 11:52, Duane Maxwell wrote:
There's actually a lot of interesting stuff in the file system we were defining. Basically, we extended Apple's dual-fork files with n-fork files, indexed with four-character codes (Apple's OSType). The conventional file is the 'data' fork, which is what you get if you simply open the file. The libraries could be files with a 'libs' fork, with extended ELF segments (in order to include fetch information) indexed by MD5.
Windows/NT's NTFS file system supports multiple data streams, as well. However, I don't know that anyone has done anything with them (other than storing Apple resource forks on them, of course). We had to provide for them in the design of the Microsoft Tape Format. You'd think that if they were so useful, people would use them. (I don't know whether the Win32 API gives any access to them, though; that may get in the way of widespread use, since many people don't like to use the undocumented Nt layer).
Of course, Windows still hasn't grasped the idea of "file type" (as distinct from "file extension") yet. They could use one of the forks to store file type, creator, etc. data (as, I believe, OS/2 did).