From: Johnny Billquist <bqt at softjar.se> Could you expand on how it fails? Duh, sorry!! Here's the most egregious case: Test of autowrap, mixing control and print characters. The left/right margins should have letters in order: L l M m mN n O o P p Q q qR r S s T t U u uV v W w X x Y y yZ z Push <RETURN> (This is a screen copy/paste from E11 but it looks identical on the real VT100.) I've apparently overwritten the log file that I'd marked up with what the problem was, but IIRC it had to do with moving the cursor around in between writes to the last column of the screen. My testing eons ago showed that on a real VT100, writing printing characters is this simple: - if hidden "wrap" flag is set and in final column and [?7h in effect, <CRLF> - if in final column, set "wrap" flag for next time (otherwise clear it) - write character, advancing cursor unless in final column (Plus the real VT100 has a bug if autowrap happens at the bottom of the screen when the scroll region ends higher up, where the cursor does the right thing but the wrapped character appears at column 80 instead of column 1 -- but that can't be intentional and it's fixed on the later terminals.) ESC 7 / ESC 8 save/restore the "wrap" flag but other than that it survives most things. But the VTTEST output seems to expect that other things you do will clear the flag. The source code I found (maybe not official?) isn't particularly commented so it's hard to tell which parts are supposed to be clever tests and which are generic screen writes. John Wilson D Bit