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

Re: [hackers-il] a simple linux registry

Expand Messages
  • Tzafrir Cohen
    ... BTW: that page is not currently availbe for some reason. Try http://sf.net/projects/registry ... Ugly, stupid, pointless: maybe. Huge: sure it isn t. Look
    Message 1 of 7 , Apr 28, 2004
    • 0 Attachment
      On Mon, Apr 26, 2004 at 02:00:37PM +0300, Elad Efrat wrote:
      >
      > On Sun, 11 Apr 2004, Tzafrir Cohen wrote:
      >
      > > http://registry.sourceforge.net/

      BTW: that page is not currently availbe for some reason. Try
      http://sf.net/projects/registry

      > [...]
      >
      > i think it's gay ugly stupid pointless messy huge and a waste of time

      Ugly, stupid, pointless: maybe.

      Huge: sure it isn't. Look at the size of the code of the library. Comare
      to gnome's gconf .

      Oh and: Why do you choose to waste your breath and time on it?

      Cheers.

      --
      Tzafrir Cohen +---------------------------+
      http://www.technion.ac.il/~tzafrir/ |vim is a mutt's best friend|
      mailto:tzafrir@... +---------------------------+
    • Beni Cherniavsky
      ... If it becomes a problem, ReiserFS is definitely a solution. Enabling this sort of things is the purpose that initially motivated ReiserFS. ... I find the
      Message 2 of 7 , Apr 30, 2004
      • 0 Attachment
        Tzafrir Cohen wrote on 2004-04-11:
        > I saw an interesting project lately: a Perfect Configuration System
        > [tm], but designed with simplicity in mind: Each key resides in its own
        > text file, and the registry sits in its own tree structure.
        >
        > http://registry.sourceforge.net/
        >
        > The only disatvantages I can think of is the disk space consumption.
        > I wonder How much can ReiserFS help here? XFS?
        >
        If it becomes a problem, ReiserFS is definitely a solution. Enabling
        this sort of things is the purpose that initially motivated ReiserFS.

        > But the atvantages are obvious: everything remains a file:
        >
        > * can use whaever permission scheme is used on the system
        > * moving a subtree is an atomic operation.
        > * I wonder if symlinks are sensible
        > * Can be easily manipulated with the tools we all know
        >
        > Comments? Flames? Will you use it in your own project?
        >
        I find the key format to be suboptimal.

        - What's that about 20..39 for binary formats and 40..254 for text
        formats? Why?!? Why not use something extensible, like names?
        Why not use something established, like MIME types?

        - I don't think a standard header would do anybody good. Without a
        header, existing file types could be used with full support from
        existing tools (think images - how many image viewers skip everything
        until ``<DATA>``?).

        - Without a header, we don't need the `rg` command. `ls`, `cat`,
        etc., SHOULD be directly useful on the "registry", otherwise you
        lose half the benefits.

        Let's see what it takes to get rid of the header:

        - Format version - not relevant if there is no registry format.

        - Data type & comments - if you need these at all, it makes more
        sense to use a gconf-like model where the expected type and
        key meaning documentation are stored separately as a "scheme".
        That way a user overwriting the key won't lose the documentation.
        That's good.

        - If you do need the metadata per-instance of the key, the ideal
        option would be to use subfiles of the file, as will be possible
        in reiserfs4 (any file also behaves as a directory). Otherwise,
        less beautiful arrangements are possible...

        That's it - no headers! No headers => no need for a special tool and
        to a great extent even no need for a library!


        --
        Beni Cherniavsky <cben@...>
        Note: I can only read email on week-ends...
      Your message has been successfully submitted and would be delivered to recipients shortly.