TODO: xml-split - cannot decode entities, first character might become < - or, could prefix a space to text data - code as state machine in C? contradicts the intention to be simple try puredata IDEA: good file formats: - text - lines - TSV - escaped data, not quotes - trailing tab - key: value blocks - length [type] block - sequence of fixed-size structures other reasonable file formats: - CSV, with "quoting" and " -> "" - key="value" (like shell script) - XML? I don't like it much! escaping: - escape: nul tab lf cr backslash -> \0 \t \n \r \\ - or use entities to encode nul, tab, lf, cr -> � - and we also need to use & if the input was plain text