> How do you think about to improve the replacement specification a bit more > by combining the variants A till D into a single SmPL rule which will use > a nested SmPL disjunction?
I haven't actually looked at the semantic patch code, but I would point out that making a disjunction is not always a good idea. The semantics of a disjunction is that if an earlier pattern matches, then the later patterns are not considered. The semantics of a succession of rules is that later patterns are applied to the result of earlier patterns. These are clearly not the same things. At the expression level, ensuring that subsequent patterns don't apply if earlier ones do entails quite a lot of work, so a disjunction may not be very efficient. This performance issue is much less at the statement level. If there are no semantics or performance issues involved, then I would be inclined to think that the way the person first decided to write the rule is the best way, because that is how they thought about the problem. julia _______________________________________________ Cocci mailing list [email protected] https://systeme.lip6.fr/mailman/listinfo/cocci
