It’s the “path of traversal” on a graph
FWIW, not all file systems are trees
Yes. Just throw all your files into the same place. Tree structures are so 90s.
Yes. I have completely forgotten about symbolic links while writing this
I always thought because it leads to a destination
And home is where the path starts
Or OP is right because root is at the base of the tree
well you see you build your tree home on a branch that comes off the root.
And the branches have webs
and bugs that get caught in your nets.
And every file and folder is a parent, child, sibling or orphan
Is there a file system graph that’s not a tree? Like a folder that contains a folder that contains itself?
symbolic links come to mind. there’s soft and hard symbolic links. soft links you probably know, it’s just one item in the tree that points to another. hard links are more intricate because you (the user) cannot easily differentiate them from an ordinary file / directory. it’s basically hacking the file system to lead you in circles. this way, a directory can contain itself, which means that it’s not a tree anymore.
Which is probably why Unix tends to disallow hardlinks to directories.
They’re not physically impossible; on Mac, Time Machine even creates them (there’s some kind of special thing it has to do to be allowed to do that, I think). The system design people just went “yeah, that’s a bad idea”.
Which is slightly impressive coming from the people who decided newlines in filenames were perfectly fine, on a system where it’s very common practice to sling data around as a list of lines.
(I like the fact that literally every other character is allowed. Including spaces. Dealing with spaces isn’t too terrible! It’d just be way easier to deal with things in a guaranteed correct fashion if newlines in filenames weren’t allowed.)
– Frost
I’ve talked before about my old idea about a tag based file system but it is purely theoretical yet afaik. Basically it is an object database where it doesn’t matter how you order the relevant tags. /a/b/file and /b/a/file are the same thing.
See the Wikipedia link in another comment about semantic filing systems seems like it is real.
Thanks!
Starting in around 2004, a new wave of implementations centered on manual tagging of files and folders.
Wow, this was about the time I came up with the idea. Then again it was the time of blogs and tag clouds, so…
I wonder is you could do some fun stuff with symbolic links and Russell’s paradox.
Does object storage count? Those are usually identified by a single key; although I guess you could count cloud tenancy and buckets as “folders”
Wouldn’t same be the case for normal files systems? I think, technically, files on disk stored as flat structure plus tree structure which tracks physical files bytes location on the disk. Is there a file system which actually stores dirs/files data as tree? Adding something into folder would mean moving data of next folder to create “room” for new stuff
Semantic file systems are very much not tree-shaped. The idea of a semantic file system is that files don’t have a single parent or container, like a folder in a traditional hierarchical file system, and instead a file’s location is represented by a combination of tags. As expected, you don’t traverse a path to reach a file, but rather, you perform a query not unlike operators in a search engine.
