jchem_core_pkg.get_hit_count problem in 5.3

User c2ffbfa8f8

02-03-2010 18:49:18

Hi,

We have recently upgraded the JChem cartridge to 5.3 and the behaviour of some estimation queries seems to have changed. If running something like:

SELECT jchem_core_pkg.get_hit_count(SCHEMA', 'TABLE, 'ISOMER', 'N=C=O', 't:na') FROM
dual

It returns:

ORA-29532: Java call terminated by uncaught Java exception:
oracle.jdbc.driver.OracleSQLException: ORA-00942: table or view does not exist
ORA-06512: at "JCHEM.JCHEM_CORE_PKG", line 193

However, if you run another query in the same session, eg:

SELECT DISTINCT ISOMER AS smiles
FROM SCHEMA.TABLE WHERE jc_compare( ISOMER, 'N=C=O', 't:s earlyResults:100' ) = 1

Then run the first jchem_core_pkg.get_hit_count() query again, then it will work fine. This doesn't seem to be any kind of issue with the SQL clients.

Seems very odd, I'm pretty certain the hit_count queries were working fine individually before the upgrade, any ideas?

thanks muchly

derek

ChemAxon aa7c50abf8

03-03-2010 10:24:34

Hi Derek,

I am not able to reproduce the problem with a simple test case: I am starting up the JChem Server and execute the hit count function before executing any structure search function/operator. Please, could you review the log of my test (attached) to see what changes I might have to make to it in order to reproduce the problem?

Also, could you, please, post the Java stack trace (to be found in the Oracle session trace file) corresponding to the ORA-00942 error?

(I assume the missing single quotation marks in your get_hit_count call are typos only in your post, as Oracle would reject the call already in the parsing phase.)

Thanks

Peter

User c2ffbfa8f8

03-03-2010 13:48:53

Hi Peter,

Thanks again for the prompt reply. Yes, the missing quotes are a typo (sorry!). Attached is a modified test case with its output, and also the stack trace from the jserver logs (hopefully these traces will be from this particular problem but I dont know 100% for sure as I couldnt retrieve them myself). The attached test case output shows an identical query failing then succeeding. I notice we're using Oracle 10g and Java 6, which is different from your test case. Please let me know if you need any further details.

All the best

Derek

ChemAxon aa7c50abf8

03-03-2010 17:23:20

Derek,

On Oracle 10.2, I could, indeed, reproduce the problem with JChem 5.3.0. This will be solved in the next minor version (scheduled for release end March).

Peter

User c2ffbfa8f8

03-03-2010 17:52:33

Thanks Peter, much appreciated.

Derek

User 9f6f294e9f

07-04-2010 10:47:44

Hi Peter

Do you have any idea when the next minor release that will include the hit count bug fix will be released ?

Regards

Ant

ChemAxon aa7c50abf8

07-04-2010 13:45:03

Hi Ant,

The current estimation is that it will be released in a few days (this week).

Regards,

Peter

User 9f6f294e9f

15-04-2010 10:39:51

pkovacs wrote:

Hi Ant,

The current estimation is that it will be released in a few days (this week).

Regards,

Peter

Hi Peter

I guess that something is still holding 5.3.2 up. Can you tell me when it will be released please ?

Regards

Ant

ChemAxon aa7c50abf8

15-04-2010 13:58:13

Hi Ant,

There is only a slim chance that it will be released tomorrow. Most likely it won't be released until next week.

In the meantime, you can download a release candidate from here. This release candidate has a buglet which requires to "manually" create the jchem properties table for new users before they can create JChem indices. This is automatically done when you connect to the new user's schema with JChemManager. Users already having JChem indices aren't affected.

Regards,

Peter