use Perl

• re-Normalize (Score:1)

  by n1vux (1492)

  I think you've overnormalized. You don't need task_start, task_finish, task_cancel junction tables. You've just traded trying to enforce UserID NOT NULL IF Status NOT NULL for enforcing that a task appears at most once in the UNION of the three task_{start,finish,cancel} tables, which will be even more awkward to code in your data dictionary (1=COUNT(*) FROM UNION...)

  You need just one task_status table that has NOT NULL FKs to TaskID and UserID, plus a State enumerated type and an autoinsert date timesta

  --
  Bill
  # I had a sig when sigs were cool
  use Sig;
  • Re:re-Normalize (Score:2)

    by Ovid (2709)

    Ah, thanks. You're correct that my normalization wasn't correct. I can now see the value of what you propose and that does make the task simpler and more maintainable. As a caveat, though, some of what you suggest implies using a real database, not a toy like MySQL (which is what we're using).

    This gives me another good rule of thumb to keep in mind. Naturally, when I see things like field_1, field_2, field_3 and so on, it's obvious that I need another table as what I'm looking at is essentially an arr
    • Re:re-Normalize / Toys (Score:1)

      by n1vux (1492)

      As a caveat, though, some of what you suggest implies using a real database, not a toy like MySQL (which is what we're using).

      Can you upgrade your MySQL to the latest & greatest? Yes, MySQL3 with default engine MyISAM is a toy. MySQL5.[01] with InnoDB engine (or BDB, or the coming Falcon engine) is no longer a toy -- subquery, views, transactions. (They've even got a cluster engine for partitioning.) [ Engines [mysql.com]] (I think this is the point of the Perl 2 joke.)

      Barbie is correct that normalizing is hard

      --
      Bill
      # I had a sig when sigs were cool
      use Sig;
      • Re:re-Normalize / Toys (Score:1)

        by chromatic (983)

        Bill nails it. I stopped having any sympathy for people who refuse to upgrade to new versions of free software at approximately 7 am on Sunday morning.
        • What happened at 7am Sunday? (Score:1)

          by n1vux (1492)

          Bill nails it. I stopped having any sympathy for people who refuse to upgrade to new versions of free software at approximately 7 am on Sunday morning.

          What happened Sunday? I don't see a journal entry for Sunday. Is there a new entry for stupid customer tricks?

          Memo to self: update home servers to Ubuntu to 6.06LTS and Debian to 3.1 now that they're really stable ...

          --
          Bill
          # I had a sig when sigs were cool
          use Sig;
          • Re:What happened at 7am Sunday? (Score:1)

            by chromatic (983) on 2006.07.05 20:08 (#48613) Homepage Journal

            I read yet another Nicholas Clark "This isn't fun anymore, and no one seems to want to pay someone to fix Perl 5" message on p5p. That's when I decided that companies which rely on Perl 5 but don't send bug reports or test snapshots and have all of their code in the Darkpan can fix their own bugs, or at least pay for the continued development of Perl. That is, they can do that if they care about their code.

            Reply to This

            Parent

            • Paying for Support for Perl5 (Score:1)

              by n1vux (1492)

              "companies which rely on Perl 5 but don't send bug reports or test snapshots and have all of their code in the Darkpan can fix their own bugs, or at least pay for the continued development of Perl. That is, they can do that if they care about their code."

              That hits home.

              Support: I tried to get commercial support for Perl, and couldn't. Yes, I could pay for it, but I couldn't get anything I could recognize as support. I talked to ActiveState at LinuxWorld 2005, interested in buying support fo

              --
              Bill
              # I had a sig when sigs were cool
              use Sig;