3 ms·

The thing I've never been able to get past with Winer's tech is that he is so married to hierarchical representation. I've always felt like my own thoughts are

The thing I've never been able to get past with Winer's tech is that he is so married to hierarchical representation. I've always felt like my own thoughts are best expressed as graphs. edit: I mean CS graphs; many-to-many; a child having multiple parents. I don't mean your basic mind-mapping graphical software, which very often also don't support representing graph data structures.

jerf 12y ago
Many outliners support graphs... look for the "clone" feature, though it's called various things in various outliners. The problem is that you have the same sort of problem you do with UNIX filesystems, where almost everything expects to work on trees, and then, suddenly, bam it's not a tree anymore and something a little funky or unexpected happens because some code expecting a tree is suddenly being run on a full graph.
- spc476 12y ago
  From "UNIX Implemenation", _The Bell System Technical Journal_ Vol 57, No 6, Part 2 (July-August 1978): "The file system structure allows an arbitrary, directed graph of directories with regular files linked in at arbitrary places in this graph. In fact, very early UNIX systems used such a structure. Administration of such a structure became so chaotic that later systems were restricted to a directory tree."
  - lepht 12y ago
    Symlinks sort of allow for graph-like structures in the filesystem, no?
    - jerf 12y ago
      No "sort of" and no graph-"like", it's just, yes. If you follow symlinks you can have a graph. Many filesystems also permit hard-links to form a graph, which makes it that much trickier for a program to determine if it's in a graph or tree regime; I believe many file systems also deny this. Symlinks + preventing directory hardlinks are a decent compromise in power, where things that choose to understand them can handle graph-like structures, but things that wish to treat filesystems naively (thus treating symlinks as just another file) can work on trees. Beware the "friendly" library that automatically follows symlinks in something like a scripting language, for they are throwing you to the wolves in the name of user friendliness.

jerf 12y ago

Many outliners support graphs... look for the "clone" feature, though it's called various things in various outliners. The problem is that you have the same sort of problem you do with UNIX filesystems, where almost everything expects to work on trees, and then, suddenly, bam it's not a tree anymore and something a little funky or unexpected happens because some code expecting a tree is suddenly being run on a full graph.

spc476 12y ago
From "UNIX Implemenation", _The Bell System Technical Journal_ Vol 57, No 6, Part 2 (July-August 1978): "The file system structure allows an arbitrary, directed graph of directories with regular files linked in at arbitrary places in this graph. In fact, very early UNIX systems used such a structure. Administration of such a structure became so chaotic that later systems were restricted to a directory tree."
- lepht 12y ago
  Symlinks sort of allow for graph-like structures in the filesystem, no?
  - jerf 12y ago
    No "sort of" and no graph-"like", it's just, yes. If you follow symlinks you can have a graph. Many filesystems also permit hard-links to form a graph, which makes it that much trickier for a program to determine if it's in a graph or tree regime; I believe many file systems also deny this. Symlinks + preventing directory hardlinks are a decent compromise in power, where things that choose to understand them can handle graph-like structures, but things that wish to treat filesystems naively (thus treating symlinks as just another file) can work on trees. Beware the "friendly" library that automatically follows symlinks in something like a scripting language, for they are throwing you to the wolves in the name of user friendliness.