wait for proof-goto-point to finish

[Abhishek Anand]

I am writing some IDE automation over proof general and want to implement a blocking version of proof-goto-point, so that the automation can then execute query commands on an uptodate state. I tried the following but it sometimes returns early. What is the right way to do it?

(defun proof-goto-point-and-wait () 
    (proof-goto-point)   
    (sleep-for 1) ;; even with this, on commands that sometimes take long time to process like `Require Import ...`, this function sometimes returns earlier, in a state where the side effects of processing until current point is not yet visible
    (proof-shell-wait))

[hendriktews]

proof-goto-point being non-blocking is a key feature of Proof General: Emacs stays responsive, while Coq is working in the background, which can easily take minutes.
In the automated tests we have a similar problem, there we use wait-for-coq or variants of it, see for instance ci/simple-tests/coq-test-prelude-correct.el.
Note that there is also a Proof General internal way to solve this problem. You can add specific flags or callbacks to action list items, see proof-shell-action-list-item in generic/proof-shell.el. Automatic background compilation and proof tree visualization use this to get notifications when certain commands have been completely processed.