At 09:31 PM 2/10/2008, John Docherty wrote:

I also found that the column in question had slightly different
definitions in the two tables (not possible ?) and that I could
change them to match, (not possible too I thought ?) so maybe
there is something 'corrupted' somewhere. No doubt the dream
team will let me know what I have wrong.

John,

Before you prepare a sample and send to R:Dream Team, I suggest
that you do your homework and make sure that your database is
not corrupted and you can reliably replicate the issue.

A few questions:

01. Have you checked the integrity of your database?

02. Have you checked the CHARacter SETtings of your database?

03. Are the character settings unique for each item?

    For example, you cannot use the same character for QUOTES
    and IDQUOTES.

04. Have you changed the settings for QUOTES?

05. Have you changed the settings for IDQUOTES?

06. Most importantly, are you able to replicate the scenario
    using the RRBYW14 sample database?

    If yes, then, prepare the sample with all details using
    RRBYW14 database and send it to [EMAIL PROTECTED]

I can assure you that R:Dream Team will not waste any time
to find the cause of your corrupted database, nor be able
to replicate anything if the database itself is not healthy.

Here are a few suggestions related to your database.

At the R> prompt:

    DISCONNECT
    SET MESSAGES ON
    SET ERROR MESSAGES ON
    SET MULTI OFF

    -- CONNECT the database with OWNER password, if any
    CONNECT dbname IDENTIFIED BY owner password

    -- CONNECT the database with no passwords
    CONNECT dbname

01. Check the connected database CHARacter settings, especially
    the IDQUOTES.

    At the R> prompt:

    CLS
    SHOW CHAR

    -- Notice the setting for IDQUOTES (the last item on list).

    If this setting is blank/null, make sure to set the IDQUOTES
    settings to ` (that is a single reversed quote, on the same
    key as the ~ tilde).

    SET IDQUOTES=`

02. Now create the unload file with NULL set to -0-.

    At the R> prompt:

    CLS
    SET NULL -0-
    OUTPUT newdb.ALL
    UNLOAD ALL
    OUTPUT SCREEN

    This step will create two files (newdb.ALL and newdb.LOB)

03. DISCONNECT, and then rename this database to some other
    name.

    At the R> prompt:

    DISCONNECT
    RENAME dbname.RB? dbnameBK.RB?

04. Now, let's rebuild the fresh database and see if it passes
    the integrity check.

    At the R> prompt:

    CLS
    RUN newdb.ALL

    -- Watch the activity and all messages on the screen.
    -- Don't fall asleep. You might miss an important warning.
    -- Completion time may vary based on size of the database
    -- (number of tables/records/indexes)

    If there were no warnings or error messages, you've got a
    fresh/healthy database.

    If there are any warning or -ERROR- message(s), take them
    seriously. In that case:

    -- DISCONNECT the bad database (built using RUN newdb.ALL)

    At the R> prompt:

    DISCONNECT

    -- Delete the bad database

    At the R> prompt:

    DELETE dbname.rb?
    DELETE newdb.ALL
    DELETE newdb.LOB

    -- Rename the previously saved database (Step 03.)

    RENAME dbnameBK.RB? dbname.RB?

    -- CONNECT the database

    CONNECT dbname

05. CORRECT all -ERROR-s accordingly and then repeat Step 02.

06. Do not quit or give up until you see a completely fresh
    database without any warnings or errors.

Hope that helps!

Very Best R:egards,

Razzak.


Reply via email to