Github user ottobackwards commented on the issue:
    I think that the assumptions that it makes about the ability to cache the 
output, and apply to every transformation or every function may be a bit 
    I would have it off by default but put it in the tuning guide, with 
something written up about the circumstance where it will work and the 
circumstances where it will not.
    A setting the marks the return as cacheable  in the configuration would be 
the better option, on a per transformation basis.


Reply via email to