Comments for sam's do  

….and… i think cursor == rs with lazy loading

Looks interesting! But the problem is to remember about that ;) To be serious, it’s a good solution.

Yes it is, but there are result sets with eager/lazy loading

Looks like some debug code made it in accidentally.

I would suggest always wrapping specs in a pending block rather than using this type of code. A pending_if helper is pretty easy to make, and an example can be found in dkubb/dm-core’s spec/lib folder. It will allow you to be notified when a spec starts passing.

Is a scrollable result set like a cursor?

You may want to remove this print statement :)

@myabc: you should always rebase against the edge prior to pushing local changes to prevent merge commits from happening. I sometimes like to do “git pull —rebase” after each commit, and re-run the specs to make sure everything works good.

I’m doing it as we speak. ;-)

Found the need for other projects to check for both location of mingw32 and host for configure options.

Will do the announcement soon.

I think that would be pretty helpful.

Else I’ll just have to add the few lines by hand to check a few locations.

Oops, that was me :-P

I found that in OSX MinGW from ports gets installed in a different place.

I’m thinking ExtensionTask should provide a helper for it, wdyt?

Would row_count and field_count be part of the public, semipublic or private APIs? You may want to document it with an @api tag like we do in dm-core.

Public means that the end user would use and depend on the method name, return values and side effects. Semipublic means that only other plugins outside of what you control use it. Private means that only code you control uses it. With the first two you should maintain some stability, but with private you are free to change/remove the method in any way.

According to the section on Compliance in the JDBC 4.0 API Specification, implementing java.sql.Connection#prepareStatement(String, int) is optional.

This is issue can be tracked as JRUBY-3286

@luislavena Yep! I sent you an email!

Hey Alex, what about making Rake::JavaJarTask for rake-compiler? ;-)

Isn’t this a JRuby bug then? I think we should properly investigate this, because this absolute style is usually a bit more reliable when people tinker with the load paths etc.

wheee!!

oooops. correct creation command: CREATE TABLE PEOPLE

hello, I looked at insert…returning problem and seems like it’s partially supported only in the newests pg drivers: http://jdbc.postgresql.org/changes.html#version_8.4dev-700 There are few ways to get generated key in JDBC (tested using last stable JDBC4 driver) i.e.: final String queryCreate = “CREATE TABLE PEOPLE”; final String queryInsert = “INSERT INTO PEOPLE (name) VALUES returning *”; //solution 1 stat = connection.createStatement(); ResultSet rs = stat.executeQuery(queryInsert); rs.next(); System.out.println("" + rs.getInt(1)); // solution 2 prepStat = connection.prepareStatement(queryInsert); prepStat.execute(); rs = prepStat.getResultSet(); rs.next(); System.out.println("" + rs.getInt(1)); // solution 3 prepStat = connection.prepareStatement(queryInsert); rs = prepStat.executeQuery(); rs.next(); System.out.println("" + rs.getInt(1)); -————————————————————————————————————- I can offer some time – may i fork or just clone DO repo and help with specs etc ?