Preventing the user from entering a number that is invalid for the currently selected base is complicated. It requires checking every keystroke as the user types it in, and handling deletions yourself, and (for obscure reasons) keeping track of what the user typed in by yourself.
It is a lot easier to validate the number at the time the conversion is requested.
- binary: all characters must be members of the set {'0', '1'}
- octal: all characters must be members of the set {'0', '1', '2', '3', '4', '5', '6', '7'}
- decimal: all characters must be members of the set {'0', '1', '2', '3', '4', '5', '6', '7', '8', '9'}
- hex: all characters must be members of the set {'0', '1', '2', '3', '4', '5', '6', '7', '8', '9', 'A', 'B', 'C', 'D', 'E', 'F', 'a', 'b', 'c', 'd', 'e', 'f'}
To prevent the user from entering invalid characters, you should study Jan's code at
Notice how he handles one character at a time, and keeps track of what has been typed, and manages deletions "manually" because MATLAB will not track them automatically when you use keypressfcn callbacks.
A more thorough way to handle the situation, at risk of portability, is discussed by Yair at https://undocumentedmatlab.com/blog/editbox-data-input-validation . He shows how to go in at the Java level and hook in validation functions to control what the user can type and what is displayed when the user does type.
