Besides having many different names (orthographic panorama - orthoscopic panorama - slit-scan panorama - strip panorama - multi-viewpoint panorama - multi-perspective panorama - pushbroom panorama) feasibility of orthopanos depends of the subject
: a painting, a flat wall, one or several buildings where the photographer can stay at the same distance from the facade/facades (but then if roofs and chimneys are visible or if there are parked cars in between difficult to hide errors will necessarily happen...)
: any situation where objects located at different distance from the camera are visible from one shot to the next. An example is a crossroad when shooting a series of well aligned building facades.
Just look through the viewfinder while moving from one shooting location to the next : you will see why this is often a difficult, very difficult or impossible kind of panorama. Another useful example comes from viewing a landscape through the window of a moving train coach: foreground is moving fast while background don't moves.
Practice, trial and error, post-processing, multi-layers are the way to go.
Free as in "sell your soul to Satan"
Anyway this academic/academic-like paper http://grail.cs.washington.edu/projects/multipano/
explains the matter better than any other I managed to read...