Consider this test_syntax.xml mode file: """ <?xml version="1.0"?> <!DOCTYPE MODE SYSTEM "xmode.dtd"> <MODE> <RULES> <SEQ TYPE="KEYWORD1">one</SEQ> <IMPORT DELEGATE="DELEGATE_1" /> </RULES> <RULES SET="DELEGATE_1"> <SEQ TYPE="KEYWORD2">2two</SEQ> <IMPORT DELEGATE="DELEGATE_2" /> </RULES> <RULES SET="DELEGATE_2"> <SEQ TYPE="KEYWORD4">three</SEQ> <SEQ TYPE="KEYWORD4">3three</SEQ> </RULES> </MODE> """ And this file handled by it: """ // Test jEdit syntax modes one two2two 3three """ The problem is, that "three" and "3three" do not get highlighted/recognized, unless you add a rule into the DELEGATE_1 set starting with the same char/number, e.g. changing "2two" to "32two" would highlight "3three" from DELEGATE_2..! I've tried to call ParserRuleSet::resolveImports() recursively at the beginning for the rulesets it processes, but this can cause infinite loops. If IMPORTS of "infinite" depth wouldn't be supported, that would make sense, but only rules with already existing hash/index chars get applied seems to be a bug somewhere. ((( Use case: I want to use this for the php.xml mode file, so that I import a list of regular expressions for function definitions into a rule set, where the regular KEYWORDs are defined, so that the keywords get highlighted as supposed, but the regular expressions (one for each hash char) are recognized also, not overriding the highlighting of the KEYWORDs.)))