Leonard Hecker

Fix various KKP encoding issues (#20052) NOTE: This commit changes our input model to treat all Ctrl-Alt combinations as an AltGr modifier by default. Ctrl+Alt+/ will not produce `"\x1b\x1f"` anymore. This fixes the following issues: * Base-key report of non-ASCII keys was broken * AltGr on international keyboards would spam Ctrl/Alt up/down events repeatedly forever * KKP doesn't report modifier key repeats * AltGr+Key combinations were reported as Ctrl+Alt+key but should've been just plain Key-But-Modified Closes #19977 ## Validation Steps Performed This cannot be easily tested in unit tests, due to the dependence on international keyboard layouts. It appears to work correctly from what I can tell.

Fix several Present issues in AtlasEngine (#20723) * D3D uses deferred execution where render commands are collected as descriptors in a ring buffer before they're sent off to the GPU. When you don't call `Flush()` (or `Present()` which implies `Flush()`) it'll fill up 100s of MB of memory rapidly. This PR solves it in two ways: * Call `Flush()` whenever we can't call `Present()` * Make calling `Flush()` unnecessary by checking if we need to render anything in the first place. * This immediately results in an issue: Not rendering means the frame latency waitable object yields immediately (because we can't wait on it twice). The solution is to wait 16ms before trying again. * If the display is off, DWM will not deque frames. `Present1()` will indicate this (`DXGI_STATUS_OCCLUDED`) but it's not a failure so we don't handle it. Because of this we'll fill the queue to its maximum (64 frames). Once the display is on again, DWM will deque at V-Sync rate, but since we render as fast as the frame latency waitable object, also signaled at V-Sync rate, we'll push frames at the same rate they're dequeued, causing a 64-frame latency (~1.1s at 60Hz). `SetMaximumFrameLatency(1)` doesn't change this. The solution is that we don't limit waiting on the frame latency waitable object to just 100ms, but `INFINITE`. * Since that fix causes the renderer to wait potentially forever, a mechanism to cancel waits on shutdown is necessary. This also uncovered a flaw in the current design, wherein it blocks the UI thread for up to 100ms every time you close or move a tab on a low frame rate (or no active) display. Closes #20226 Closes #20342 Closes #20716 ## Validation Steps Performed * Continuously render, scroll up -> no GPU use ✅

Fix visual artifacts when using QQ Wubi IME (#20542)

Avoid window events during ASB (#20617) Important context: To make a window smaller programmatically, the win32 console APIs have since forever required you to * narrow the viewport * then narrow the buffer To make the window larger you had to * enlarge the buffer * then enlarge the viewport This is because a viewport had to fit into the buffer and vice versa. ConEmu was very popular in the past and used this to implement proper terminals around conhost. When WSL came along this resulted in a problem: It wants SIGWINCH-style notifications. And there are none on Windows. A long time ago, a hack was introduced that simply raised buffer size events on any viewport size changes. The problem: It was raised for _any_ viewport size changes. And it was also raised for ASB switching. Problema numero due: OpenSSH for WIndows is like WSL, but it has a severe bug: A SIGWINCH during a partial input read will corrupt the read and the other side will read either garbage or duplicate input. Well, we can fix that by just not raising buffer size events. It's very dangerous, because it may easily regress ConEmu (etc.). See: https://github.com/PowerShell/Win32-OpenSSH/issues/2275 Also: Fun fact, we've been ignoring `ENABLE_WINDOW_INPUT` since Windows 1607! No way fixing this breaks anything! /s So, this PR introduces two **regressions**: * `ENABLE_WINDOW_INPUT` is being respected again * `WINDOW_BUFFER_SIZE_EVENT` is not sent anymore when the viewport changes if VT mode disabled Changes: * Reintroduce support for `ENABLE_WINDOW_INPUT` * Coalesce buffer size events * Avoid such events when switching ASB on/off * Ensure we reliably raise the events for viewport size changes This is part of #281. ## Validation Steps Performed * Spawn OpenConsole * Run `ConEmuC64.exe -AUTOATTACH -GHWND=NEW` * Run `wsl -- bash` * Run `while stty size; do read -rt .05 ||:; done` * Narrowing the window reduces the WSL width ✅ * Widening the window increases the WSL width ✅

Fix two KKP ReportEventTypes issues (#20551) * Dead key releases should not be reported * Modifier key releases should only be reported if `ReportAllKeysAsEscapeCodes` is also set * (Also fixes encoding for numpad Enter) Closes #20522 ## Validation Steps Performed * Switch to Swiss French (`0000100c`) layout * Enable modes 0b11 (`DisambiguateEscapeCodes`, `ReportEventTypes`) * Pressing `Shift+]` produces `!` ✅ * Pressing `] ] ] A` produces `¨¨ä` ✅

Fix 5s lags when using OSC 52 clipboard copy (#20728) Hilarious repro: `printf "\e]0;A\a\e]52;c;Qg==\a"` This issue is exactly what happens when you Ctrl+X in a clean file in msedit and it freezes Windows Terminal. It's because it puts a "•" in the title and copies the text. It triggers a title and clipboard change in the same frame which causes the UI and VT thread to deadlock trying to get the console lock.

Do not retry a DSR-CPR if the ConPTY host did not respond (#20510) People sometimes use ConPTY just for console API translation and so the host lacks any VT capabilities. This means DSR-CPR queries never get a response. Previously we would never clear the dirty flag which would result in a 500ms lag on each `GetConsoleScreenBufferInfo` call and PowerShell calls that API _**on every single stdout print**_ (sometimes per character due to lack of any stdout buffering). ## Validation Steps Performed * Lobotomize conhost's ability to respond to DSR-CPR * Run OpenConsole --> VtPipeTerm --> pwsh * Resize the window * After this PR, stdio will not lag

Introduce an AI Usage Policy (#20654) After quite a lot of deliberation today, I think this is how I would personally define it. I tried to keep it short and open to AI-use, but particularly restrictive to autonomous AI-use (= typically slop).

ConPTY: Do not wait for DA1 on startup (#20536) A counterpart to #20510. After this commit you can spawn ConPTY again without handling any queries and it'll (mostly) just work.

Fix KKP auto-repeat events (#20500) Turns out, ReportEventTypes does not control autorepeat. Closes #20499