In many ways, Transitdocs reminds me of a situation I was in 30 years ago...that of being 'downstream' from the data-supplying system.
While working as a contractor at a Baby Bell, I was given the project of writing an extract program to get all changed data from the 'Ma Bell' CRIS customer billing system IMS hierachical data base and use it to populate their shiny new DB2 relational data base for marketing. Each data base held 30-40 billion data records (rows) so the extract had to run as fast as possible every night just to 'keep up'. The problem was that the team that handled the billing system wanted nothing to do with the marketing team I was on. So, what happened too frequently was they'd change a record structure or add something new that the extract program was totally 'unaware of'. I'd be 'aware' of the problem with a 2-3AM phone call from operations indicating that 'my' extract program 'blew up' ('blue screen in Windows terms) and I had no choice but to immediately go in to work and fix it.
The exact same 'downstream' issues occur with the Youtube download program I use. Youtube changes their 'protocol' or <whatever> and the next time I try to extract a video, it dies and I have to download and install an updated version of the extract program.
Transitdocs is in the same boat. Every time Amtrak changes something on their site, the only 'learn' of it when their data retrieval activities cease to function. Sometimes it can be fixed in a couple of minutes, sometimes hours, and sometimes longer. It depends on how complex the changes made by Amtrak are.
Now if I could only get Amtrak to update their Guest Rewards database instantaneously when my ticket is scanned. UPS, Fedex, and even the US Postal Service scans their packages as they are delivered and their computers are instantly updated. Why not Amtrak?