Re: [PATCH] jakarta-turbine-torque ( was: RE: [PATCH] Sybase testbed )

[Byron Foster <bf...@base2.cc>]

On Thu, 2001-12-13 at 21:13, Brekke, Jeff wrote:
> 
> First off, thanks for the patch/work Byron.  I've been wanting to get around
> to verifing the decoupled torque with sybase, but we're in deep with turbine
> 2.x at the moment.  Anyway, after your patch was posted, I munged around
> with it.
> 
> We went around with the delimiters when the current version of torque was
> updated to support sybase.  Sybase's isql tool can also define the delimiter
> to use to terminate a command ( its a command-line switch ), then there's
> always sqsh, or java alternatives.  I think it is simpler to have the same
> delimiter within torque, which is really just using the same delimiter that
> the sql task in ant was using.  So I'm -1 on switching the sybase templates
> or the adapter class to use 'go' as the deilimter

All makes sense.  I didn't realize you could still import the sql files
with a ';' as a delimiter through isql.  Given that, I agree that the
issues simplifies sticking with the same delimiter.

> Your patch to the TableTool in TorqueDataDump worked for me with sybase, but
> mysql hung.  I took the easiest way out which was to re-implement that inner
> class using Village.  This effectivly reads the entire result set into a
> QueryDataSet which allowed the Iterator interface to be implemented
> correctly.  Not much different but has some trade offs.  The first being the
> entire dataset is read into memory: large tables will suck.  The other is
> since Torque is currently implemented with Village, the data dump task will
> work with any jdbc driver that village ( and subsequently torque ) works
> with.

Strange it didn't work with mysql. I tested it with mysql before I sent
it in. Are you using the mm.mysql-2.0.7-bin.jar?  I had to change
TableTool because it was using the ResultSet.isLast() method which is
not supported by jConnect (Sybase seems a little behind in JDBC 2.0
compliance).  However, it sounds like the change you made has the best
chance of working with the assortment of JDBC drivers.  What are your
thoughts about memory usage when exporting data?  Loading all the data
into the ResultSet could be a strain.  
 

> Could you, or anyone else, try this out?  I've tested it with sybase and
> mysql and all seems OK here.  Thanks.
> 
> jb
> 

I tested it with sybase and mysql and both ran ok.  
+1 for the patch.

Byron



--
To unsubscribe, e-mail:   <ma...@jakarta.apache.org>
For additional commands, e-mail: <ma...@jakarta.apache.org>