Now that the InputHandler thread blocks with select, in addition to the
main thread that also blocks with select, winpty has hit a bug in the
original MSYS where select returns EAGAIN. select is not supposed to fail
with that error (EAGAIN doesn't make much sense as a select error). Add
a workaround just for the original MSYS.
Explicitly shut down the I/O handling threads before restoring the terminal
state. Start calling winpty_close.
This change doesn't fix any bug, AFAIK. It's nice to test/demonstrate the
use of winpty_close.
The InputHandler now uses select to wait on both the blocking stdin tty and
a wakeup fd. This is the same behavior I saw in OpenSSH. OpenSSH sets
its select FDs to O_NONBLOCK *except* FDs for which isatty returns true.
It puts tty FDs into raw mode, but it does not set VMIN/VTIME to 0/0, so
the terminal driver blocks. (In any case, I tested the 0/0 values, and
IIRC they prevented blocking only with a Cygwin emulated pty -- console FDs
in Cygwin blocked; as did both consoles and emulated ptys in the original,
old MSYS. O_NONBLOCK prevented blocking in all cases, though.)
Terminals output these codes under various circumstances:
* rxvt-unicode: numpad, hold down SHIFT
* rxvt: numpad, by default
* xterm: numpad, after enabling app-mode using DECPAM (`ESC =`). xterm
generates `ESC O <mod> <letter>` for modified numpad presses,
necessitating kBareMod.
* mintty: by combining Ctrl with various keys such as '1' or ','.
Handling those keys is difficult, because mintty is generating the
same sequence for Ctrl-1 and Ctrl-NumPadEnd -- should the virtualKey
be '1' or VK_HOME?
The intent is to make it easier to determine how Windows creates KEY_EVENT
records. The code lives in winpty-agent because (1) it's very small, and
(2) I can't test all the interesting keyboards/locales/etc, so if
someone hits a bug, placing it in winpty-agent.exe ensures that they'll
have access to the tools needed to debug it.
* I don't anticipate caring about this, but in principle, there's nothing
wrong with a NUL character in the input map, so fix this while the code
is fresh in my mind.
* Remove these characters from the input map and handle them via the
ordinary <Char> and Alt-<Char> code paths.
* Refactor lookupKey to remove the isEof parameter.
* Ctrl-2 (^@) is encoded as a NUL byte.
* Instead of looking for a NUL byte, we need to check the input size.
Places affected: scanKeyPress, matchDsr, lookupKey
* N.B.: putty and rxvt encode Ctrl-Alt-2 as `ESC NUL`.
Previously, after inputting a NUL using Ctrl-2, one could type an arbitrary
amount of input into the input buffer, and winpty wouldn't do anything
until 1-second inactivity timeout had expired. (i.e. It thought it was
seeing an incomplete DSR, so it kept waiting for the DSR to complete.)
* ^[O is Shift-Alt-O, not Alt-O. Alt-O is ^[o
* '[' is not a valid virtualkey. It's typically VK_OEM_4 on a US
keyboard, but it needs to be looked up dynamically.
* The unicodeChar fields should be 'O' and '[', not NUL.
* Add a --showkey argument to the unix adapter.
My Ubuntu machine has a nice showkey -a utility that dumps keys, but I
can't find something similar for Cygwin/MSYS. It's a very useful tool
for debugging UNIX terminal interoperability, so add it to the unix
adapter.
* Add WINPTY_DEBUG=input to log input bytes and console key events.
* Remove GIT_REVPARSE_WORKS, which is never used
* Rename the shell variables {UNIX,MINGW}_GPP to {UNIX,MINGW}_CXX to
match the config.mk variable names.
There is a "version suffix" that defaults to "-dev". A maintainer could
change the suffix (or remove it) by invoking make:
make BUILD_SUFFIX=-foo
make BUILD_SUFFIX=
It can also be changed in gyp builds:
gyp winpty.gyp --depth=1 -D BUILD_SUFFIX=-foo
gyp winpty.gyp --depth=1 -D BUILD_SUFFIX=
If git cannot be executed, the string "none" is used for the commit hash.
* This change reduces the total build time from about 14 seconds to about
9 seconds on my computer.
Also:
* Consolidate all the intermediate files into the build directory to
reduces clutter.
* Allow specifying UNIX_ADAPTER_EXE to make. Perhaps this will be helpful
to the MSYS2 fork, which renames console.exe to winpty.exe. (I like
the renaming, but I don't know about the other winpty users. Maybe
I'll make the change after I've put out another stable release...)
* Rename the WINPTY define to COMPILING_WINPTY_DLL define. The longer
name is clearer. I define the macro inside libwinpty/winpty.cc, so the
build system no longer needs to. (I removed the define from
winpty.gyp.)
* Consolidate config-unix.mk and config-mingw.mk into config.mk. The
separation was previously necessary because each file had a conflicting
definition of CXX.
* Rename the UNIX_LDFLAGS_STATIC_LIBSTDCXX macro to UNIX_LDFLAGS_STATIC,
because libstdc++ isn't the only thing I'm linking statically.
I think it might be possible for m_dirtyLineCount to be greater than the
windowRect.top() + windowRect.height(). If it is, then scanForDirtyLines
will perform an invalid read while trying to initialize prevLineAttr.
Also add agent/EventLoop.h. I don't know why it wasn't listed.
With these changes, gyp no longer complains about missing files.
Unfortunately I can't do more testing at the moment.
This commit might fix issue https://github.com/rprichard/winpty/issues/55.
