Rather than building the timestamp capability directly into every persistence layer supported by AASM, could the timestamp mechanism generate a callback under the covers that calls the setter for the column?

This will also keep more advanced logic out of every persistence layer so that they can focus purely on persistence.

@reidmorrison comment sounds good to me, @CapCap @Drenmi What do you guys think ?

Sounds like a sensible approach @reidmorrison. 🙂

@reidmorrison I like it too. Not sure how much time I have to do it though, but will try

what callback should I use to set timestamp column?

before is ok