Warning: I know zero about Java :-) I am trying to connect (read only) to an IBM Cloudscape database. This is what eventually became Derby. I have a list of jar files: db2j.jar db2jcc.jar db2jcview.jar db2jnet.jar db2jtools.jar that a real Java application uses to access the same database. The database is in a local directory like: find db -type f -print | head -20 db/OpcClientDb/db.lck db/OpcClientDb/dbex.lck db/OpcClientDb/log/log.ctrl db/OpcClientDb/log/log48978.dat db/OpcClientDb/log/logmirror.ctrl db/OpcClientDb/seg0/A1 db/OpcClientDb/seg0/c10.dat db/OpcClientDb/seg0/c101.dat db/OpcClientDb/seg0/c111.dat db/OpcClientDb/seg0/c121.dat db/OpcClientDb/seg0/c130.dat db/OpcClientDb/seg0/c141.dat db/OpcClientDb/seg0/c150.dat ... Poking at the internet so far I've some up with this: ==== #! /usr/bin/env jruby include Java require 'lib/db2j.jar' require 'lib/db2jcc.jar' require 'lib/db2jcview.jar' require 'lib/db2jnet.jar' require 'lib/db2jtools.jar' url = 'jdbc:db2j:db/OpcClientDb' conn = java.sql.DriverManager.get_connection(url) ==== which gets me: DriverManager.java:602:in `getConnection': java.sql.SQLException: No suitable driver found for jdbc:db2j:db/OpcClientDb (plus a stack trace) So I am guessing that I need to load the classes in the jar files but I can't figure out how to do that. I've tried things like: import "com.ibm.db2j" or include_class java::db2j or java_import java.lang.com.ibm.db2j all of which don't seem to be even close to what I need. If I do: jar tf lib/db2j.jar | head -20 I get things like: com/ibm/db2j/aggregates/AggregateDefinition.class com/ibm/db2j/aggregates/Aggregator.class com/ibm/db2j/authentication/UserAuthenticator.class com/ibm/db2j/catalog/AliasInfo.class com/ibm/db2j/catalog/DefaultInfo.class com/ibm/db2j/catalog/DependableFinder.class com/ibm/db2j/catalog/IndexDescriptor.class com/ibm/db2j/catalog/ReferencedColumns.class com/ibm/db2j/catalog/Statistics.class com/ibm/db2j/catalog/TypeDescriptor.class com/ibm/db2j/core/BootStrap.class How do I figure out what to pass to import or java_import or include? Thank you for your time, Perry
on 2013-05-17 03:50
on 2013-05-17 04:20
Are these java classes view by the jre thats runs Jruby? Sounds like a classpath problem... El 17/05/2013 03:50, "Perry Smith" <firstname.lastname@example.org> escribi:
on 2013-05-17 07:08
Hi Perry, you need to load the driver class (right) before connecting using DriverManager. Assuming the class name is right do smt like: Java::JavaClass.for_name 'com.ibm.db2j.jdbc.DB2jDriver' Make sure you do this after the .jar files are required, yet before connecting ... Hope it helps, K.
on 2013-05-17 15:18
Weee!!! Thank you so much. I now get: WARNING: Cloudscape (instance c013800d-013e-b298-9d99-00c0a801c300) is attempting to boot the database /Users/pedzan/Source/OPCstuff/ruby-stuff/db/OpcClientDb even though Cloudscape (instance c013800d-013d-b6e5-23bc-0009318f4f00) may still be active. Only one instance of Cloudscape should boot a database at a time. Severe and non-recoverable corruption can result and may have already occurred. But I am pretty sure that is because they create file locks and I have a snapshot of the DB while its open. Now, I just need to figure out how to dump what I need but thats probably a topic for another list. By the way (for the next guy who might follow). I did: jar tf lib/db2j.jar | grep Driver to get the exact name (which Karol had predicted correctly) -- and then changed the slashes into dots. The thing I don't 100% understand is why not just do this for all the classes when the jar file is loaded? Would that be a huge performance hit or something? Thank you very much! Perry
on 2013-05-17 15:56
please note that `jar | grep` only works if the driver author named their java.sql.Driver implementation with "Driver" in the class name. since you mentioned 0% Java knowledge you can not expect to 100% understand this :) - it's a Java (actually JDBC) thing ... scanning (and thus loading) all .jar classes for driver implementations (which might have arbitrary names as mentioned) would certainly be a performance hit - there's a Java auto-discovery convention (META-INF/java.sql.Driver text file inside the .jar) that gets around this it's likely that your driver does not conform with it thus the manual loading. enjoy Java with JRuby - it's certainly more enjoyable this way !
on 2013-05-17 20:34
On top of Karol's mentioning of performance reasons...There is no nice API for reflectively querying all classes loaded. Jar loading possibly could be done via examining classpath (although classpath can change at runtime), but JVM can also dynamically generate classes on the fly (or even get loaded by JNI). The ability to just auto-process all defined things is difficult even if you ignore the performance penalty of processing thousands of classes you probably will never consume. -Tom
on 2013-05-24 11:40
The following does not directly help and its unsolicited but if you use Ruby Sequel which I am currently using via jdbc drivers to hook up to Oracle and Postgres all of your load issues should go away. Furthermore, the ruby sequel dsl is nice to work with but you can also just pass it raw sql or a combination of. Disclaimer I have not tried this out with Cloudscape but if there's a jdbc driver for it , my understanding is that it should work. May need to drop that jar under the jruby libs directory. Check it out, you might be happier. Charles
on 2013-05-24 14:00
Thanks. I will definitely check it out.