Coq execution difference between semicolon “;” and period “.”

Question

Given a valid Coq proof using the ; tactical, is there a general formula for converting it to a valid equivalent proof with . substituted for ;

Answer 1

The semantics of tac1 ; tac2 is to run tac1 and then run tac2 on all the subgoals created by tac1. So you may face a variety of cases:

There are no goals left after running tac1

If there are no goals left after running tac1 then tac2 is never run and Coq simply silently succeeds. For instance, in this first derivation we have a useless ; intros at the end of the (valid) proof:

Goal forall (A : Prop), A -> (A /\ A /\ A /\ A /\ A).
intros ; repeat split ; assumption ; intros.
Qed.

If we isolate it, then we get an Error: No such goal. because we are trying to run a tactics when there is nothing to prove!

Goal forall (A : Prop), A -> (A /\ A /\ A /\ A /\ A).
intros ; repeat split ; assumption.
intros. (* Error! *)

There is exactly one goal left after running tac1.

If there is precisely one goal left after running tac1 then tac1 ; tac2 behaves a bit like tac1. tac2. The main difference is that if tac2 fails then so does the whole of tac1 ; tac2 because the sequence of two tactics is seen as a unit that can either succeed as a whole or fail as a whole. But if tac2 succeeds, then it's pretty much equivalent.

E.g. the following proof is a valid one:

Goal forall (A : Prop), A -> (A /\ A /\ A /\ A /\ A).
intros.
repeat split ; assumption.
Qed.

Running tac1 generates more than one goal.

Finally, if multiple goals are generated by running tac1 then tac2 is applied to all of the generated subgoals. In our running example, we can observe that if we cut off the sequence of tactics after repeat split then we have 5 goals on our hands. Which means that we need to copy / paste assumption five times to replicate the proof given earlier using ;:

Goal forall (A : Prop), A -> (A /\ A /\ A /\ A /\ A).
intros ; repeat split.
 assumption.
 assumption.
 assumption.
 assumption.
 assumption.
Qed.