I’ve got a language, whose compiler is able to output detailed information about the grammar of currently parsed file. Would it be possible to use this information instead of the
tmLanguage (or any other regex-based gramar description format) to provide the grammar for Atom’s highlighter?
Compilers are able to provide much better description than regex-based tools (sometimes based on inferred types, or available or missing symbols) and the work don’t need to be duplicated. Moreover, such solution would be much faster, because Atom won’t need to run many regexps matches over the text, it just gets information about meaning of each part of source code.