I will have a very heavy bag. And be travelling with a gentleman who's not as nimble on his feet as he once was, also with luggage. So we will change at Cheltenham Spa. Why on earth would the journey planner advise anyone to change at Gloucester? I can't see the advantage in so doing.
Shortest-path routing engines are my speciality rather than multi-modal timetable-based ones but...
Given two different journeys producing identical outcomes, the route chosen will generally be a random, implementation-dependent pick from the two. So the trick is to weight the outcomes. WP has mentioned one way this can be done (better score for long layovers) but it should be equally possible to downrate a transfer that involves struggling over a long footbridge.
Some journey-planner software does support "preferred transfer" points, places where changing trains incurs a smaller penalty. Here's the relevant ticket for OpenTripPlanner, for example:
https://github.com/opentripplanner/OpenTripPlanner/issues/365In this case, it depends whether the software commonly used in
UK▸ rail sites supports it (I'd be surprised if it doesn't) and whether the people putting the data in have defined Cheltenham Spa as a preferred transfer point.