Hello, I am looking at the parsing of org-table formula mode strings and I find it quite confusing and I have an hard time understanding if it works the way it does by design or by accident.
Formula mode strings are documented here https://orgmode.org/manual/Formula-syntax-for-Calc.html#Formula-syntax-for-Calc but the specification is not very strict and the parser implements surprising behaviors. Would it make sense to tighten the specification and the implementation? I think being a bit more strict is what is accepted would help catching typos and mistakes, but I am not sure that having formulas resulting in #ERROR upon an org-mode upgrade is desirable. Also, maybe a specific marker (#INVALID ?) should be used for invalid syntax. The main problem with the current parsing code is that any character that is not recognized as a valid mode flag is used as a value format string. For example: | 1 | 2 | OO3.000 | #+TBLFM: $3=$1+$2;FOO%.3f Is this by design? If so, what is the use case? Thus is slightly more surprising: | 1 | 2 | 3.000 OO | #+TBLFM: $3=$1+$2;%.3f FOO I argue the first form should result in an error. The second can either result in an error or in everything following the % sign to be used as a format string, although I am not sure there is a clear use case for this. Thank you. Cheers, Dan
