On Tue, May 27, 2014 at 8:42 AM, Mathias Doenitz <[email protected]>wrote:
> Ideally, since PEG parsers can potentially backtrack all the way to very > beginning of the input, in these cases the parser should also return an > info about what prefix (length) of the total input can now be dropped, > because the grammar will never backtrack beyond it. I've found that if any part of the semantic processing (like translation) is performed after parsing, one may need access to any part of the input to have good error reporting. But if one was to discard input, it could be where Mizushima et al identified as places for removal of memoization information with the honoring of "cut" operators. -- Juancarlo *Añez*
_______________________________________________ PEG mailing list [email protected] https://lists.csail.mit.edu/mailman/listinfo/peg
