I have been reviewing a paper where the authors apply a computational method from an earlier paper they published to a specific system of general relevance. The paper seemed ok, and I recommended major revision in the review.
After submitting the review, I've been working with this computational method myself and discovered that the way the authors applied the method is seriously flawed, so that it makes in general no sense to interpret the result in any way without further tests. If I had noted this problem earlier, I would probably have recommended rejection in my review.
What's the recommended course of action in this situation? Should I notify the editor of this additional discovery, so that the authors can directly take care of it in their (presumably ongoing) revision, or is it better to just wait for a request to review the revised version?
I haven't heard any decision from the journal so far, but I don't think that the paper will be accepted without revisions.