On 18.05.2012 23:59, Y M wrote:
Approaches tried: to create an encoder that holds a level-layout, later a level-encoder map, and choose the appropriate object inside 'doEncode'- the question is that some initialization is needed on layouts and encoders, including a context on both situations, and sharing this context between all these objects does not seem to work, not sure why. Using a map of encoders I tried to use them separately for encoding sharing the same context, no good-- with a map of layouts I tried to replace my encoder's layout with the appropriate one, then back to the original layout, and it also does not work. I'm not even posting my code here, as I believe this is the wrong way to implement this.
Also, just now reading the archive, I noticed that the previous messages, from Shane Kelly, are basically the same issue as mine with a different approach. Hopefully a solution can fit both.
I think is the easiest approach would be to use evaluators within the ThrowableConverter. All the supporting code is already there. Cheers, -- Ceki http://twitter.com/#!/ceki