On Thu, May 15, 2008 at 2:58 AM, <[EMAIL PROTECTED]> wrote: > > Why would that be the wrong library? isofs uses diskfs, and isofs does > > a similar thing: takes a file and shows it in another way, as a > > directory tree. Isn't the filtering translator supposed to do the same > > thing: take the underlying directory and show it filtered? > > What isofs does is offering access to the filesystem structures (inodes > etc.) contained within a store. Thus it uses libdiskfs -- a library for > "real" filesystems that directly store the filesystem data in the > underlying store. (Usually a disk partition. It can also be a normal > file, but it will still be treated like a "real" store -- just like > loopback devices on other systems, only less awkward...) >
Does it mean that the store should be a single file (special file)? I thought that the store could be a directory, too. So, if I were implementing a tar file translator, I should have used libdiskfs, do I get it right? > > Unfourtunately, I cannot even suppose what another way to write such a > > translator could be :-( > > For virtual filesystems, where the filesystem structures are mapped to > something else rather than stored directly, you should use libnetfs. > (Which is a misnomer: libvirtfs or something like that would fit much > better -- just like libdiskfs should rather be named libstorefs...) Frankly speaking, I didn't even suppose that libNETfs could be used for this purposes :-) Okay, I'm happy I've acquainted myself with libdiskfs (thought quite superificially), and I'm diving into ftpfs now. I hope I'll be able to cope with new information faster this time. I also hope it is not really bad that I lost time playing with libdiskfs... scolobb
