Loading ...
Sorry, an error occurred while loading the content.

Re: Sorting the keys of formatted (pretty printed) JSON output

Expand Messages
  • kriszyp
    ... when ... Map.Entry. ... O(N). ... Not true. Full rehashes are indeed O(N), but they take place at a frequency proportional to 1/c because each time a
    Message 1 of 7 , Apr 10, 2008
    • 0 Attachment
      --- In json@yahoogroups.com, John Cowan <cowan@...> wrote:
      >
      > Tatu Saloranta scripsit:
      >
      > > > There would be a penalty for keeping a TreeMap versus a HashMap.
      > > > HashMap obviously performs better (constant time put & get) for
      when
      > >
      > > I understand this, and it's a valid point.
      >
      > Actually, it isn't. Even though the javadoc claims that get and put
      > are constant time, a moment's reflection shows that this cannot be so.
      >
      > HashMap is implemented (in OpenJDK, and almost certainly in earlier
      > versions) using an array of specialized objects that implement
      Map.Entry.
      > Each element of the array is one hash bucket, and the Entry objects form
      > a linked list. So average get/put time is approximately N/c*2, where c
      > is the capacity (initially 16 unless specified otherwise), which is
      O(N).
      > The hope is that N/c*2 is always small enough that the performance looks
      > like O(1), but as N grows, c is increased using a full rehash, which is
      > again O(N) and is independent of c.

      Not true. Full rehashes are indeed O(N), but they take place at a
      frequency proportional to 1/c because each time a rehash takes place
      the size is doubled, so the rehashes become farther apart the larger c
      gets. Therefore the average rehash time for a put operation is
      O(N*1/c) which averages to constant time. Both get and put only have
      to iterate through their own bucket, so as long as the hash method is
      reasonable, both get and put are indeed O(1), constant time operations.
      Kris
    • Douglas Crockford
      ... Excellent suggestion. I have modified JSON.Object.
      Message 2 of 7 , Apr 10, 2008
      • 0 Attachment
        --- In json@yahoogroups.com, "Arthur Blake" <arthur.blake@...> wrote:
        > When I have machine generated JSON data that is stored in files, I
        > like to have the JSON data be sorted alphabetically by the keys in the
        > output.
        >
        > It's much easier to find what you are looking for quickly this way
        > when viewing or editing the data (especially when the amount of data
        > is large.)
        >
        > Because there is no defined order of the keys in JSON, it doesn't
        > matter what order you put them in. So we might as well put it in an
        > order that makes it more human readable in contexts where the data is
        > meant to be viewed or edited by humans.

        Excellent suggestion. I have modified JSON.Object.
      Your message has been successfully submitted and would be delivered to recipients shortly.