I see. So octave expects to invoke "info" with certain options, and alternatives that prove infobrowser don't interpret those options correctly. And some packages that provide: info-browser (ahem, emacs, ahem) don't even provide an infobrowser executable alternative.
You're absolutely doing the right thing, but it seems to me that there's opportunity for improvement if packages providing info-browser were required to provide an infobrowser alternative executable and these were required to accept a short list of options. This would allow octave to present its info in emacs, say, via whatever shim mechanism emacs sets up to interpret those options and spawn an emacs or emacsclient. I'm not sure where a wishlist bug like that would be appropriately filed, but I suspect the emacen folks would be the right place for the debian emacs support. Any ideas? And more importantly, does the whole idea sound sensible to you? Cheers, --Barak.