The question is on how the "agent of infinitive" slot may be filled for this syntactic structure: with a form of direct object, or indirect object, or either.
[Agent of verb of perception] + [agent of infinitive] + [verb of perception] + [object of infinitive] + [infinitive]
One set of examples come from this Web page.
Je les entends le casser
(I hear them breaking it)
Je le regarde l'écrire
(I'm watching him write it)
Je le sens la caresser
(I feel it caressing it)
--in which the slot is filled with what is ordinarily called a "direct object"--les and le. (Never mind whether they are direct objects here).
The other example is from La porte étroite by Andre Gide, as follows:
Dès le vestibule, j’aperçus Alissa ; elle semblait m’attendre et vint aussitôt vers moi. Elle portait au cou, dans l’échancrure de son corsage clair, une ancienne petite croix d’améthyste que je lui avais donnée en souvenir de ma mère, mais que je ne lui avais pas encore vu mettre. Ses traits étaient tirés et l’expression douloureuse de son visage me fit mal.
--in which the slot is filled with an "indirect object"--lui. (Again, maybe not an indirect object here.)
I believe this lui is filling the same syntactic slot as les and le in the other examples. For example, les are heard but do the breaking. In the same way, lui is seen but does the wearing (mettre). Therefore, I can only see these possibilities.
The top three examples are wrong, and the Gide clause is right.
The three examples are right, and Gide is wrong.
All four are correct, and each has the only form of pronoun it can for the agent of infinitive. There is a principle to explain why the top three must have les and le (and not leur or lui) while Gide must have lui (and not la).
All four are correct, and you get a choice in this type of context and can go with either le etc. (the "direct" form) or lui etc. ("indirect"). In this case, there might be other principles to explain why one form or another is preferable.
The question is:
Which of the above four possibilities is right?
If 3, please also explain what the principle is.
If 4, please say something about whether one form or another is preferable and why.
This question is related to this other from me, which I believe was rather poorly formulated. If anyone sees a logistical issue, I will be happy to delete the old question as repetitive, or merge this one into that, or whatever is appropriate. Thank you.