This is a list of known problems in gawk 3.0. Hopefully they will all be fixed in the next major release of gawk. Please keep in mind that the code is still undergoing significant evolution. 1. The interactions with the lexer and yyerror need reworking. It is possible to get line numbers that are one line off if --compat or --posix is true and either `nextfile' or `delete array' are used. Really the whole lexical analysis stuff needs reworking.