On 08/27/13 22:34, Jonathan Wilkes wrote:
Conclusion: teach Fanout(1) and Trigger(2) for situations where ordering doesn't matter, and Trigger(1) for situations where it does. The end.
only that many Fanout(2) problems originate in a Fanout(1) design, where at some point the patch was extended and branches where execution order did not matter suddenly are merged again in a way where execution order *does* matter.
i think that most patchers have heard about [trigger] and it's merits, it just doesn't occur to them that in their specific patch execution order does matter. i dare say that most of the buggy patches posted here (and elsewhere) are buggy exactly because a Fanout(1) mutated into a Fanout(2).
as for simon's asynchronous semantics of fan-out, it's probably better to start using it only *after* it has been implemented.
conclusion: always make execution order explicit, even if you currently don't care about it. the end.
fgmadrs IOhannes