CompuServe Thread

The original design criterion for file name lookup (Tripos) was that most of the time, the full file name was known… so speed up that kind of open. Hashing the directory makes a lot of sense in that case. For all of us developers (read hackers) who can't remember the time of day, let alone the names of the files we've been working with, however, "dir" "list" and "whereis :abc#?" are very common commands….. ….and we wait …….and wait ………..and finally something appears on the screen. Well, I guess there are tradeoffs everywhere. Yeah, I wish the "dir" command were faster, but I guess I should trade some of the development time for speedier applications. Regarding a "flat" UNIX file structure…..not any that I've seen….. they are all hierarchial these days. UNIX made them famous, MS-DOS followed suit….I think even DRI has finally gone to it. BTW Tripos predates UNIX. Regarding file recovery—I wish the programs to do it were better, but disksalv seems almost perfect. try stepping on the inode that defines the root directory in a UNIX system sometime, then recover the disk. One last thing in defense of those at Cambridge (the place where Tripos was invented). They use 32bit pointers for the sector id's….I don't currently see or envision any time that I will need 4 giga-sectors, but most UNIX systems can't handle even 4 giga-byte files (I don't have a need for them either…..but) don't forget your MS-DOS couldn't handle any volume larger than 32 mega-bytes until recently. It all boils down to what you really "need" out of a system. Enough flaming….. as someone once said "It's a poor workman that blames his tools." We all need to learn how to use the tools we have effectively. Just because they are not the ones we learned doesn't mean they are "wrong". That point of view is extremely provincial. Everything we learned at "Mommy's knee" ain't neccessarily so. Ok, Ok, Ok……I'll quit beating the dead (I hope) horse.