Closed upstreammuse closed 2 years ago
With the move to a GUI version, this issue is basically moot. From here out, file errors need to be one of two cases:
It's worth pointing out that there are almost certainly cases where type (2) errors are being treated as type (1) errors, but this is honestly not a high priority to fix right now, since a file-manipulating user is expected to have a workflow that revolves around updating the file directly based on feedback from the program (in which case the file read error popups do the job), and a GUI user won't get into these problems to begin with.
There should be a validation layer between reading the file (syntax validation) and semantic validation. The semantic validation is necessary because a human made the input file, and should still be used as part of reading a gui-saved file. However, the semantic validation isn't related to processing errors, so it could be in a layer before the real processing starts. Essentially there should be a distinction between the kind of validation that needs to run because a human wrote an input file that could be internally inconsistent from the kind of validation that always needs to run to check for balance issues, savings issues, and logical things that don't come from bad file input.