The main rule in java mode makes \ as escape character for all rules. This is incorrect because \ should work as escape only inside String and character literals. Example: \/** does not start javadoc comment even though it obviously should. The same applies of course to \"foo" The escape should be removed from RULESET and literal SPANs should be defined as follows: " [^\\]" ' [^\\]' The problem with this is that SPAN_REGEXP treats only the begin sequence as regex. Also the HASH_CHAR attribute is put there for performance reasons. I have nothing against good performance but optimizations should be applied ONLY when they are possible and do NOT compromise correctness. Thus HASH_CHAR should be optional and used only if the regex starts with a specific character.

I searched for existence of terminate rule in jEdit's mode files and found that it was used only in text mode to turn off parsing.

The help files say that it is used in patch mode which is not true and obviously should not be used if the intention is to highlight the whole lines.

Could someone give an example in what syntax could one use Terminate rule?

I had a good look at gnu.regexp package that jEdit uses for regex matching. It does seem like a good package. The customizability is better than in java.util.regex. But where java is little better is the documentation. In gnu.regexp one can tweak the syntax to one's liking but many of the variations are not explained. And definitions of different syntaxes one has to finally dig from the source.

This is exemplified with jEdit's syntax which was PERL5 syntax with character classes enabled and line separator set to \n.

What still puzzles me is that in jEdit documentation \< is said to match at the start of a word. But gnu.regexp documentation says this belongs only to Egrep syntax.