Book I section 13.2 gives us a good look at impersonal gerundives and I will go over what Goodwin says compared to how Thucydides uses them.
πρῶτοι δὲ Κορίνθιοι λέγονται ἐγγύτατα τοῦ νῦν τρόπου μεταχειρίσαι τὰ περὶ τὰς ναῦς, καὶ τριήρεις ἐν Κορίνθῳ πρῶτον τῆς Ἑλλάδος ναυπηγηθῆναι.
So the bolded words are imperfective eventive impersonal gerundives. The first is executive voice and the second is passive voice.
In our aspectual paradigm, these are substitutes for conjugated verbs. Further, the second one is an intransitive structure.
Note that the first one has a logical subject. It’s not a grammatical subject, despite being in the -oi case; it’s not the subject of legontai, “they say”, which is in base voice, not executive voice. The three English translations on Perseus, and the Smith translation for the Loeb Classics Library, agree that the Korinthioi didn’t say whatever it was. Legontai is an idiom for other people saying something.
So it is the Korinthians who metakheirisai’d, deliberately to bring about having ships. When Goodwin says that the subject of an “infinitive” is in the accusative, he’s wrong.
Smyth allows as how the subject of an infinitive can be in the -oi case but he has requirements, none of which apply here:
1) The subject of the infinitive is also the subject of a conjugated verb in the same sentence. We don’t have any conjugated verbs here.
2) The i.g. is substantivized with an article in an oblique case (not -oi). The definite articles in this sentence have noun complements; the i.g.s are not substituting for nouns, they are substituting for conjugated verbs.
Smyth deals with this subject in section 1973 on page 439; his discussion of the -ous subject of an i.g. is in section 936 on page 260. This is one of the many pitfalls of the old grammars: segregating information that should be addressed in the same place, even if that means repeating information. And, as I said, neither of these conditions applies to 13.2.
The other i.g. is passive voice and triireis is the grammatical subject and logical object.