>>>>> "Tom" == Tom Tromey <[EMAIL PROTECTED]> writes:
Roman> Look out for a line with 'BUG' in the scanner Roman> output. Tom> Oh, duh, there it is. Sorry :( Ok, this patch fixes the bug for me. I'm checking it in. I see a few other instances of this problem in the code. In general this lookahead scheme is a bit flawed, since the code often pushes back a 'char' ... but for proper EOF handling it should push back the raw int. Another solution here would be to check read() results more uniformly. So much for my earlier theory :) Tom 2006-08-18 Tom Tromey <[EMAIL PROTECTED]> * gnu/javax/swing/text/html/css/CSSScanner.java (readWhitespace): Push the 'int', not the cast char. Index: gnu/javax/swing/text/html/css/CSSScanner.java =================================================================== RCS file: /cvsroot/classpath/classpath/gnu/javax/swing/text/html/css/CSSScanner.java,v retrieving revision 1.1 diff -u -r1.1 CSSScanner.java --- gnu/javax/swing/text/html/css/CSSScanner.java 16 Aug 2006 21:31:27 -0000 1.1 +++ gnu/javax/swing/text/html/css/CSSScanner.java 18 Aug 2006 20:10:22 -0000 @@ -602,7 +602,7 @@ chc = (char) ch; } // Push back last character read. - lookahead[0] = chc; + lookahead[0] = ch; }