Summary from today's hacking meeting:

Order of the evaluation

• The adjust rules would have to be defined in the right order so that the added context dimension is defined early enough during the adjust rules application (single fmf node can have definition at 3 locations thanks to elasticity)
• There could be several run-throughs, during the first pass we could just evaluate rules which are modifying the context dimensions
  • This would not work if the expectation is that the changes should be applied to selected fmf subtrees only
  • Changing context for follow-up rules seems to be good enough, no expectation of multiple passes

The original use case seems to be ok with change in the current set of rules only

from fmf.