On Sat, Mar 2, 2013 at 5:19 PM, Andy Seaborne <[email protected]> wrote:
> On 02/03/13 10:02, Claude Warren wrote:
>
>> First some background.
>>
>> I am debugging an app and did kill the app in the middle of an execution.
>>
>> <code>
>> dataset = TDBFactory.createDataset( "./jena" );
>> model = dataset.getDefaultModel();
>> </code>
>>
>> the getDefaultModel() method throws a null pointer.
>>
>> in DatasetPrefixesTDB.**readPrefixMap()
>>
>> <code>
>> public synchronized Map<String, String> readPrefixMap(String graphName)
>> {
>> Node g = Node.createURI(graphName) ;
>> Iterator<Tuple<Node>> iter = nodeTupleTable.find(g, null, null) ;
>> Map<String, String> map = new HashMap<String, String>() ;
>> for ( ; iter.hasNext() ; )
>> {
>> Tuple<Node> t = iter.next();
>> String prefix = t.get(1).**getLiteralLexicalForm() ;
>> String uri = t.get(2).getURI() ;
>> map.put(prefix, uri) ;
>> }
>> Iter.close(iter) ;
>> return map ;
>> }
>> </code>
>>
>> graphName = "";
>> t = [<null>, <null>, <null> ]
>>
>> String prefix = t.get(1).**getLiteralLexicalForm() ;
>> throws a null pointer exception.
>>
>> My question is: What can I do to restore the data? I tried reloading
>> using
>> tdbloader2 to reload the data but that did not work.
>>
>
> Did you delete the old database? tdbloader2 is (currently) not
> prefix-aware so it may be that the old prefix table is being left in place.
> Try in a fresh directory.
>
> You can also just go in and delete the prefix table files, prefix*.
>
> An empty prefix table will be recreated on the next start up.
>
>
> My other question is where should I look to find the problem?
>>
>
> At guess, there was some non-transactional use of the database. A change
> to the prefix table did not get completely written (the node file changes
> were not written but the tuples part was).
>
> This could be due to an interrupted load or use of TDB non-transactionally.
>
> Andy
>
>
Your last line was spot on. It was an interrupted use of TDB
non-transactionally. The only change was that I added namespace prefixes.
I ended up removing the prefix table files and that recovered nicely.
Should TDB detect a null namespace triple and delete it, or perhaps delete
the namespace prefixes at that point? Is this an enhancement that makes
sense?
--
I like: Like Like - The likeliest place on the web<http://like-like.xenei.com>
Identity: https://www.identify.nu/[email protected]
LinkedIn: http://www.linkedin.com/in/claudewarren
