On Sat, May 08, 2021 at 06:53:14AM -0700, Loren Wilton wrote: > Henrik (or anyone) what happens if a net rule is fired but no response is > ever received? > > I assume there is some timeout limit for net responses to show up, so that > message processing can complete? > > If that is true (there is a timeout on responses) then the situation in bug > 4549 still exists to the extent that there could be un-run meta rules that > were dependencies of the responses that never arrived. Per my offhand > processing proposal in bug 7735, these metas would simply not be run.
Nothing ambiguous here. As shown in both bugs, there are many scenarios where a rule is simply not run or finished. > But Daniel makes a good point in bug 4549 that these incomplete metas could, > instead of being discarded, be run twice with both true and false (or 0 and > 1) substituted for the missing net rule result, and if the meta results are > the same for both flavors the meta rule could be completed with that result. > That could result in running other incomplete metas. > > I don't honestly know if the extra processing would be in any way > worthwhile, and I can easily construct a scenario where it might be better > to not run such a meta. But I can also believe that in the majority of cases > it could be worth trying such a strategy. Since metas are (at least > currently) only boolean logic and arithmetic, they are cheap to run. And if the meta is depending on multiple unfinished rules, or even other metas with unfinished rules? Sounds like a logic nightmare.. better just design the metas better in the first place..
