From: p DOT dalgaard AT kubism DOT ku DOT dk (Peter Dalgaard BSA) Subject: Hints on the accenting (dead-key) bug 4 Apr 1997 16:31:27 -0800 Approved: cygnus DOT gnu-win32 AT cygnus DOT com Distribution: cygnus Message-ID: Original-To: gnu-win32 AT cygnus DOT com X-Mailer: Gnus v5.3/Emacs 19.34 Original-Sender: owner-gnu-win32 AT cygnus DOT com I have previously reporteded the accent-eating behavior of GW32. It occurred to me that maybe it was getting done right in NT emacs, so I took a look at the source code and actually it does seem like it is doing something sensible in w32term.c. (This is in the ordinary emacs 19.34 sources available from all over the place). However, I have neither the setup nor the time and diskspace required to actually go in and modify things and test out a new CYGWIN32.DLL, so I thought that maybe someone else would be interested in having a peek? The offending code in the cygwin sources is somewhere in winsup/fhandler.cc (I think...) around the FakeReadFile and fhandler_console::read routines. Recall: The bug affect all accenting, not just those producing characters above 0x80 but also for instance pressing '^ ' on a DK keyboard, where the '^' is a deadkey. Normally, the hat-space sequence produces a single circumflex character, but not in GNU-WIN32. There, I get nothing, leaving me no (simple) way of typing a ^ character to bash. The problem disappears if readline is not used (as in cat > file), which leads me to suspect that the problem is in the part of FakeReadFile that gets executed if ENABLE_LINE_INPUT is false. -- O__ ---- Peter Dalgaard Blegdamsvej 3 c/ /'_ --- Dept. of Biostatistics 2200 Cph. N (*) \(*) -- University of Copenhagen Denmark Ph: (+45) 35327918 ~~~~~~~~~~ - (p DOT dalgaard AT biostat DOT ku DOT dk) FAX: (+45) 35327907 - For help on using this list, send a message to "gnu-win32-request AT cygnus DOT com" with one line of text: "help".