Roc,
Thanks for the reassurance - I think project level permissions will be good. However the recommended permission check (using limit which should be equivalent to using Exists) takes 400ms for 'the public' on this test box.
In contrast the alternative method that I am experimenting with takes just less than 2ms. That is a 200 fold improvement. I can't see why that could be a bad thing.
Surely if we are interested in scaleability we should be interested in ways to improve performance - even if the status quo is really good already! 😊
Maybe I have got it wrong and there is a flaw in what I have done (in which case I welcome the inevitable corrections!), but I do think that it deserves scrutiny - rather than see it simply rejected as unnecessary.
R.
