[Mulgara-dev] dateTime information being corrupted

[Paul Gearon]

Am I correct in assuming that you are doing an "insert-select" here?  I'm
guessing this, since the object you describe is a variable.

Based on this assumption, you've already started to use logging, so can you
turn on debug logging in:
src/jar/resolver-store/java/org/mulgara/resolver/store/StatementStoreResolver.java

Line 359 does this following log:
logger.debug("Inserting statement: [" + statements.getSubject() + " "
                                                + statements.getPredicate()
+ " "
                                                + statements.getObject() +
"] in " + model);

This should tell you if the problem is the statement you retrieved (the log
will contain the wrong value).  Or the statement you insert (the log will
contain the correct value).

Paul

On 11/2/06, David Moll <DMoll at viewpointusa.com> wrote:
>
>  I'm wondering if anyone else has seen this problem.  We've been having an
> issue where a triple element of the form:
>
>  "2006-10-05T20:12:49"^^http://www.w3.org/2001/XMLSchema#dateTime
>
> Will end up changed to something else.  The change is normally subtle,
> it's not a complete corruption of the data, but simply an incorrect
> dateTime.  So instead of the above correct date we may see:
>
> "5006-10-05T20:12:49"^^http://www.w3.org/2001/XMLSchema#dateTime
>
> or
>
> "2006-11-03T20:12:49"^^http://www.w3.org/2001/XMLSchema#dateTime
>
> For instance, today (11/2/2006) we are seeing files show up that have the
> dateTime as "2006-11-03," which is clearly not right.  We are logging the
> iTQL queries that are being used to insert the triples containing this
> information.  The triples are of the form:
>
> $Subject <http://www.viewpointusa.com/vss/elements/1.0/modifiedDate>
> $Object
>
> We also have vss:createdDate, vss:updatedDate, vss:accessedDate.  $subject
> is usually just a prefix identifier with a GUID.
>
> When we find a corrupted dateTime, we can go back to the original iTQL
> query that inserted the triple into the datastore and manually run the
> insert again, which gets the correct data into TKS.  So we're not sure where
> the dateTime element is being changed.
>
> Has anyone seen anything like this happening with dateTime elements
> before?
>
> ~Dave
>
>
> _______________________________________________
> Mulgara-dev mailing list
> Mulgara-dev at mulgara.org
> http://mulgara.org/mailman/listinfo/mulgara-dev
>
>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://mulgara.org/pipermail/mulgara-dev/attachments/20061102/c02c50ef/attachment-0001.html