I would especially like an answer in the context of the theory of formal functions, although input from other perspectives is also welcomed.
The question arised when I was analyzing the main theme (or main themes?) of the first movement of the Piano Sonata nº 7 in C, k. by Mozart. In measures 7-8 we have a root position dominant moving to a root position tonic in a way that sounds to me as a cadential evasion, due to sudden change in dynamics, register and texture. What follows is a restatement of the 4 opening measures which, after a greatly extended continuation, lead to a perfect authentic cadence in m. 21. As an evaded cadence is weaker than a perfect cadence and the basic idea is clearly brought back, I'm inclined to hear this as a (compound) period. Nowhere, as far as I could find, does Caplin mention that a antecedent may end in an evaded cadence. To be sure, he does not forbid it. Hence, the question: Can an evaded cadence end the antecedent of a period?
If that is not the case, then I believe m. 8 would bring a PAC and the beginning of a new main theme (as a transition clearly starts at m. 21, after a PAC) which would leave us with two main themes with identical basic ideas and a weak/strong cadential pattern. As single main themes are the more normative option and as I am not used to considering two themes with such similar openings and with such a cadential pattern to not be a compound period, that option feels rather uncomfortable.