Given a grammar file, show the dependencies on .tokens etc... Using ST, emit a simple "make compatible" list of dependencies. For example, combined grammar T.g (no token import) generates: TParser.java : T.g T.tokens : T.g T__g : T.g For tree grammar TP with import of T.tokens: TP.g : T.tokens TP.java : TP.g If "-lib libdir" is used on command-line with -depend, then include the path like TP.g : libdir/T.tokens Pay attention to -o as well: outputdir/TParser.java : T.g So this output shows what the grammar depends on *and* what it generates. Operate on one grammar file at a time. If given a list of .g on the command-line with -depend, just emit the dependencies. The grammars may depend on each other, but the order doesn't matter. Build tools, reading in this output, will know how to organize it. This is a wee bit slow probably because the code generator has to load all of its template files in order to figure out the file extension for the generated recognizer. This code was obvious until I removed redundant "./" on front of files and had to escape spaces in filenames :(