Thanks Malte,
I hardly notice any performance leak on our implementation of ACS 3.4, in fact most of the times it only reads a cookie and does a couple of ifs, I mean, if the callback process itself isn't time consuming my guess is it won't be noticeable slower.
I will probably try it both ways and see which one works better :)
on a side note, Do you (or anybody else) know when/if that callback backport to OACS 5.1 will be available for download?