[Mulgara-dev] Mulgara Backup File -- Strings listed twice in the string pool?

[Paul Gearon]

On Tue, Mar 18, 2008 at 12:42 PM, Ben Hysell <BenH at viewpointusa.com> wrote:
>  On to our current issue, during the process of reading in the string
>  pool entries I'm getting strings being listed twice in the backup file,
>  i.e.
>
>  If early in the backup file I have
>  999  a less interesting string
>  1000 My fancy string
>  1001 some other string
>
>  Later on in the file I'll also have
>  49998 A new string
>  1000 My fancy string
>  49999 Another new string
>
>  The listing of 'My fancy string' with the same node number '1000'
>  appears twice in the string pool.  Any thoughts on how this could be
>  happening?

When you say "fancy" you're just saying that to identify the string as
being different, right?  There's no fancy unicode characters in it or
anythere, are there?

Also, are the string entries in lexical order?  I haven't checked the
backup recently, but I believe this is how things are stored.  If that
is the case, then is string for 1000 out of order?  I'm presuming that
everything is in order, but that string that appears twice is actually
out of order the second time around.

>  To throw another twist to the situation, this happened yesterday and the
>  resulting check through the Triples revealed I had gained over 1000
>  blank nodes into the system.  Yesterday we restored back to a database
>  we knew had no blank nodes in it, moved it to production, and ran for
>  the last 20 hours without incident.  This morning, no new blank nodes,
>  but at 12 noon our backup application reported multiple string pool
>  entries, but no blank nodes.  I'm not sure if I caught the system before
>  it was about to experience a blank node explosion or not...

How do you determine what is a blank node?  Do you check node IDs in
the triple list, and then use this to look up the string pool?

Paul