We study the consequences of temporal stress perturbations on earthquake nucleation in a continuous fault model. Using a two-dimensional (2-D) quasi-dynamic model of a strike-slip fault governed by a rate-and-state friction law with depth variable properties, we show that dynamic triggering (due to stress pulses or wave packets), although allowed by our results, is an exception rather than a rule and should be limited to understressed areas such as areas of high pore pressures or to faults at the very end of their earthquake cycle. When periodic stress perturbations are sensitive, the response of the fault is frequency-independent for periods lower than a period T0 but strongly depends on the frequency for periods larger than T0. We demonstrate that the crossover period T0 is equal to the time left until the earthquake instability. According to our model, high frequencies are demonstrated to have a higher triggering potential than low ones, which makes tidal triggering very unlikely before the end of the cycle due to the very low amplitudes of the stress perturbations involved.