* Instead of reading the output line-by-line, figure out what lines we
need ahead-of-time and issue as few read calls as possible. On Windows
8 and up, we issue just one read call. On earlier versions, we avoid
reading more than a certain amount.
* This change reduces the CPU usage. e.g. In my Windows 10 VM, the idle
CPU usage of winpty-agent.exe+conhost.exe combined, with an empty
console, dropped from ~3.6% to ~1.4%. In a Windows 7 VM, I measured a
reduction of CPU from ~1.6% to 0.6%.
* Increase the MAX_CONSOLE_WIDTH from 500 to 2500. The limiting factor
now is that LargeConsoleRead reads at least one line at a line, but we
don't want to read more than 2500 characters in one call on old operating
systems.
* Fix the attribute handling in scanForDirtyLines. (The assignment to
newAttr was dead.)
The 2500 limit is arbitrary and could probably be increased. The actual
hard limit depends on the OS and is around 17000. My understanding is that
the limit is based upon the need to allocate I/O buffers within a shared
64KiB heap, and I'm worried about heap fragmentation. I know that 2500
is safe, because winpty has been issuing reads of almost 3000 characters
already to find the sync marker.
Fixes https://github.com/rprichard/winpty/issues/44
* Previously, "make tests install" would also copy trivial_test.exe and
trivial_test.d to /usr/local/bin.
* The `install` tool seems to be the conventional way of installing files.
Setting permissions when copying to /usr/local/bin seems like a good
idea.
* Invoke install with -s to strip the symbol tables when they're
installed.
Instead of erasing a whole line before writing a line, simply write the
line. At the end of the line, issue a CSI 0K to erase from the cursor to
the end of the line. The erasing serves two purposes:
* It can clear characters beyond the right end of the terminal, which
have been obscured by a terminal resize.
* It sets the background color for the trailing whitespace.
Doing the erasing like so has two benefits:
* It's more correct. Previously, the trailing whitespace was colored
arbitrarily. Now, it is correctly colored to the last color on the
line.
* It should eliminate a lot of flicker.
* On output: detect them and transcode them properly into UTF-8.
* On input: create input records for both halfs of the surrogate pair.
I tested by copying-and-pasting U+20000 (D840 DC00) into a mintty bash
shell. It works, but the Windows console thinks the character occupies
four cells when it really occupies two, so the cursor position is wrong.
When I press backspace, it doesn't delete the whole character -- instead,
it replaces it with a '?', because it becomes an invalid surrogate pair.
Still, the behavior seems like an improvement.
* Use the Bold(1) SGR parameter in only one case -- White-on-Black.
* Use the 0x9X and 0x10X SGR parameters to set intense/bright colors for
both foreground and background.
Detailed changes:
* DkGray background:
- Previously, this background was handled identically to the Black
background.
- Now it is treated like the non-grayscale colors, and the
foreground/background are set exactly.
* Black background:
- LtGray foreground is unchanged: default text colors
- White foreground is unchanged: text is bolded
- DkGray is fixed: instead of concealing the text, set the foreground
to DkGray, falling back to LtGray.
* LtGray background:
- Previously, this background was handled exactly the same way as the
White background.
- Now it is treated like any other non-grayscale color.
* White background:
- Black foreground is unchanged: only an Invert is output.
- LtGray: previously, the text was concealed. Now it is handled like
Black, and only an Invert is output.
- DkGray: previously only Invert was output. Now DkGray is also
output.
- White is effectively unchanged; it is still concealed if possible.
Also: set the console colors to LtGray-on-Black on startup. The
heuristic makes little sense with other colors.
Fixes https://github.com/rprichard/winpty/issues/39
The second cell can still have a different color than the first cell, and
*maybe* we want to output it. I'm unsure. It could perhaps matter for a
character appearing at the end of a line, but it seems unlikely to matter
in practice.
Fixes https://github.com/rprichard/winpty/issues/41
* Remove the old "Vista" code path in favor of two code paths:
- setSmallFontVista: sets a TrueType font to match the code page
- setSmallFontXP: uses SetConsoleFont to activate the smallest
viable font in the console's font table
* Selecting a CJK-specific TrueType font is important so that double-width
characters occupy two cells in the console, which is important for
maintaining consistency between the console and the Unix terminal.
(Perfect consistency here might be impossible, but using the appropriate
font improves things.) winpty currently duplicates double-width
characters, but that will be fixed soon.
* MSYS is 32-bit only.
* MSYS2 and Cygwin are either 32-bit or 64-bit. On 64-bit MSYS2/Cygwin,
we use the same architecture for all generated binaries.
* Prefer mingw-w64 to mingw. mingw-w64 seems to be more up-to-date.
It is the only option for 64-bit Windows, so stop looking for
x86_64-pc-mingw32-g++.
* Also pass -static when linking binaries. Otherwise, a pthreads DLL
is linked dynamically.
* Add debug information.
* This change fixed a bug where the new winpty-agent.exe process
successfully started, but immediately exited without reaching main().
The bug occurred while SSH'ing into the Cygwin SSH server.
* Unfreeze the console while changing the buffer size. Changing the
buffer size hangs conhost.exe. See:
- https://github.com/rprichard/winpty/issues/31
- https://wpdev.uservoice.com/forums/266908-command-prompt/suggestions/9941292-conhost-exe-hangs-in-win10-if-setconsolescreenbuff
* Detect buffer size changes and switch to a "direct mode". Direct mode
makes no attempt to track incremental console changes. Instead, the
content of the current console window is printed. This mode is
intended for full-screen apps that resize the console.
* Reopen CONOUT$, which detects apps that change the active screen buffer.
Fixes https://github.com/rprichard/winpty/issues/34.
* In the scroll scraping (scrollingScrapeOutput), consider a line changed
if the new content is truncated relative to the content previously
output. Previously, we only compared against the line-buffer up to the
current console width. e.g.
If this:
|C:\Program|
turns into:
|C:\Prog|
|ram |
we previously left |C:\Program| in the line-buffer for the first line
and did not re-output the first line.
We *should* reoutput the first line at this point so that, if the line
scrolls upward, and the terminal is later expanded, we will have
output an "Erase in Line" CSI command to clear the obscured "ram" text.
We need to update the line-buffer for the sake of Windows 10 combined
with terminals like xterm and putty. On such a terminal, if the
terminal later widened, Windows 10 will restore the console to the
first state. At that point, we need to reoutput the line, because
xterm and putty do not save and restore truncated line content extending
past the current terminal width.
- Formatting a wchar_t* argument with %s does not work with MinGW's
swprintf. (Apparently it does with MSVC? Sigh.) AFAICT, %ls does
the right thing with both MinGW and MSVC. (I didn't test MSVC.)
Avoid zero-initializing the buffer to accommodate VMs that have
little RAM. Use a smaller buffer size (9000x9000 ==> 324MB
buffer) that still demonstrates the read-size limit on XP.
