Allow for expressions in Eidos event times about slim

[bhaller]

Yes, this sort of thing would be useful. There are two problems. One, script blocks get parsed and set up in the scheduler before anything gets executed, so the value of N is not known at the time that the expression would be encountered. And two, "constants" are not really constant; N can be removed with rm() and re-defined with a different value, so the expressions involving N are not truly constant expressions. Those two difficulties have prevented me from doing anything with this sort of idea, until such time as I come with a solution. However, rescheduleScriptBlock() makes it pretty easy to achieve this sort of thing anyway, so it's mostly a matter of convenience; the functionality you want is fairly easily achieved, just with a less intuitive and pretty syntax than one might wish.

[grahamgower]

Rather than using rescheduleScriptBlock(), which requires that an existing block be defined, I've been using registerLateEvent(). So the example given would become:

initialize() {

defineConstant("N", 1000);
initializeMutationRate(1e-7);

initializeMutationType("m1", 0.5, "f", 0.0);

initializeGenomicElementType("g1", m1, 1.0);

initializeGenomicElement(g1, 0, 99999);

initializeRecombinationRate(1e-8);

}


1 {
    sim.addSubpop("p1", N);
    // start outputting after burnin, for twenty generations
    sim.registerLateEvent(NULL, "{ sim.outputFull(); }", 10*N, 10*N+20);
}


For longer expressions, I create a function that is scheduled using registerLateEvent. The major disadvantage of this approach is that syntax errors in the lambda expressions don't have the appropriate source code line reported.

[bhaller]

Rather than using rescheduleScriptBlock(), which requires that an existing block be defined, I've been using registerLateEvent(). ... The major disadvantage of this approach is that syntax errors in the lambda expressions don't have the appropriate source code line reported.

Yes; with rescheduleScriptBlock() errors are correctly attributed. Error attribution within lambdas and user-defined functions is something that could use some work in Eidos...