Ok. I guess it makes sense in this case, as the encoder is invoked on
saving the model, so CM needs to access it in runtime.
Andrus
On Aug 21, 2008, at 9:29 PM, Michael Gentry wrote:
It looks like the main goal of the change was to get rid of the stack
trace printing, so I'm going to roll back and then incorporate that
change into it. Throwing the exception prevents you from loading a
model when you don't have the specified encoder in your class path.
I've just made the change and am testing now.
Thanks,
/dev/mrg
PS. If you click on the DataNode with an invalid encoder, CM gives you
an "Invalid Password Encoder" dialog stating "A valid Password Encoder
should be specified (check your CLASSPATH)." This might could be
cleaned up a little (showing it on model load), but it at least lets
you load the model and also that there is a problem with it.
On Thu, Aug 21, 2008 at 2:17 PM, Andrus Adamchik <[EMAIL PROTECTED]
> wrote:
On Aug 21, 2008, at 9:08 PM, Michael Gentry wrote:
DataSourceInfo.java changed to throw an exception if it can't load
the
password encoder instead of swallowing the exception and load the
model without the encoder present. I'll fix it.
Well, but that change was intentional. Are you going to roll it back?
