Fixes #1756 Fixes #1525 Fixes #1509 Fixes #1428 Fixes #1414 Fixes #1382 Fixes #1358 Fixes #1345 Fixes #1257 Fixes #1146 Fixes #964 Fixes #952 Fixes #210 Fixes #71

This compiles and after reviewing looks sane enough. However I am running into an issue where the text output of the terminal only shows up on the lower left quadrant of the whole window.

I figured I could go ahead and open this up as a work in progress to get some feedback and suggestions for what could have gone wrong.

This takes the latest glutin master to port Alacritty to the EventLoop 2.0 rework.

This changes a big part of the event loop handling by pushing the event loop in a separate thread from the renderer and running both in parallel.

TODO:

• [x] Resize issues (https://github.com/jwilm/alacritty/pull/2438#issuecomment-505518232)
• [x] run_return support for macOS (https://github.com/rust-windowing/winit/issues/1005)
• [x] Resizing extremely slow
• [x] Font size change not updating pty size
• [x] Message bar does not update PTY size properly
• [x] Incorrect window dimensions when moving Alacritty between different DPI monitors
• [x] Font size not updating when moving Alacritty between different DPI monitors
• [x] Clean up code / resolve remaining TODOs
• [x] Incorrect startup DPI on X11 (https://github.com/rust-windowing/winit/issues/998) -> ~~:zap: Still broken when spawning a floating window in i3~~ -> Workaround implemented
• [x] Make sure before merging all the linked issues have change log entries
• [x] Clear errors on message bar reload

Fixes #2796. Fixes #2694. Fixes #2643. Fixes #2625. Fixes #2618. Fixes #2601. Fixes #2564. Fixes #2456. Fixes #2438. Fixes #2334. Fixes #2254. Fixes #2217. Fixes #1789. Fixes #1750. Fixes #1125.

I use a window manager instead of tmux. With other terminal emulators when I want a new terminal instance, I simply open a new window. As far as I can tell, the only way to accomplish this with Alacritty is to open a new instance of the application. This is inferior for three reasons:

1. Load times when opening a new instance of Alacritty are subpar. For a program that prides itself on being the fastest, opening new terminals should be no exception. (Loading a new instance of Alacritty lasts about 1 second for me vs instant when opening a new window in iTerm.)

2. On macOS opening multiple instances of Alacritty fills up the Dock and Application Switcher.

3. Currently, such behavior must be implemented manually by the user.

From what I can tell from other issues, using window managers as a replacement for tmux is within Alacritty's scope. Even when using tmux, it can be desirable to open a second tmux instance for an unrelated session without detaching from the current session. I think multiple windows should be supported natively by Alacritty.

The Grid<T> is now a view into a region (I called it the 'active region') of a VecDeque. When new lines are added, they are pushed onto the back of the queue, and once the scrollback buffer is considered 'full', then old lines are popped off the front of the queue. This seems to work rather well since, unlike a Vec, elements don't need to be reshuffled around very often.

Of course, as discussed in #124 there are a few things that need to be fixed before I'd consider this 'ready':

• [x] configurable max scrollback length
• [ ] write new tests
• [x] fix existing ref tests
• [ ] measure performance impact (it seems fine but some numbers would be more reassuring)
• [x] fix out of bounds errors on resize

I've been looking for an extremely fast terminal emulator for many years. xterm has unfixed rendering bugs and is slower than molasses. I don't like urxvt's lack of memory management. st uses more memory than xterm (yep!).

Just saw Alacritty on HN and was determined to try it out.

What are your plans for older hardware? I'm currently using a couple of laptops from 2007-2008: a ThinkPad T43 and a ThinkPad T60. Unfortunately the Rust nightly you're currently using is broken on i686 (#126) so I can't test Alacritty on the T43, but it reported the error in the title on my T60 (which has a slightly better GPU, FWIW).

T43 specs:

OpenGL vendor string: X.Org R300 Project
OpenGL renderer string: Gallium 0.4 on ATI RV370
OpenGL version string: 2.1 Mesa 11.0.2
OpenGL shading language version string: 1.20

01:00.0 VGA compatible controller: AMD/ATI [Advanced Micro Devices, Inc.] RV370/M22 [Mobility Radeon X300]

T60 specs:

Extended renderer info (GLX_MESA_query_renderer):
    Vendor: X.Org R300 Project (0x1002)
    Device: ATI RV515 (0x7145)
    Version: 11.2.2
    Accelerated: yes
    Video memory: 128MB

01:00.0 VGA compatible controller: Advanced Micro Devices, Inc. [AMD/ATI] RV515/M54 [Mobility Radeon X1400]

Yep, I realize these are kind of the "oh wow" of GPU land :) but I'm genuinely wondering what the best approach would be to support GPUs like these - I've been pondering making my own terminal emulator for a little while (as well as a 2D engine too, actually), but have absolutely no knowledge of OpenGL, Vulkan, etc.

"It would be impossible" is completely understandable; "it would require a nontrivial architecture redesign" also wouldn't be too surprising.

For bug reports, the following information can help speed up the process. Please describe the bug that you have found and what you would expect to happen instead.

System

OS: Linux/BSD/macOS/Windows Version: alacritty --version output Linux/BSD: X11/Wayland, DE or Compositor and WM Windows: WinPTY/ConPTY (see alacritty -v output)

Logs

Crashes: STDERR, STDOUT Font/Terminal size: alacritty -vv Keyboard and bindings: alacritty --print-events

Which operating system does the issue occur on?

• NetBSD 8.99.48 NetBSD 8.99.48 (GENERIC) amd64
• Using X11 from base

While running cargo build in the checkout at commit 84aca672964e29b5b4503b7da7bc34fc395f08ab, I get this error:

   Compiling copypasta v0.6.0 (/home/ng0/code/re-src/rust_crates/alacritty/copypasta)
error[E0599]: no method named `set_urgent` found for type `window::Window` in the current scope
   --> alacritty_terminal/src/display.rs:493:29
    |
493 |                 self.window.set_urgent(is_urgent);
    |                             ^^^^^^^^^^
    | 
   ::: alacritty_terminal/src/window.rs:59:1
    |
59  | pub struct Window {
    | ----------------- method `set_urgent` not found for this
    |
    = help: items from traits can only be used if the trait is implemented and in scope
    = note: the following trait defines an item `set_urgent`, perhaps you need to implement it:
            candidate #1: `winit::os::unix::WindowExt`

error: aborting due to previous error

For more information about this error, try `rustc --explain E0599`.
error: Could not compile `alacritty_terminal`.

To learn more, run the command again with --verbose.

Implement damage tracking and reporting.

This allows compositors to only process damaged (that is, updated) regions of our window buffer, which for larger window sizes (think 4k) should significantly reduce compositing workload under compositors that support/honor it, which is good for performance and battery life.

On Wayland, clients are expected to report always report correct damage, so this makes us a good (or at least, better) citizen there. It can also aid remote desktop (waypipe, RDP, VNC, ...) and other types of screencopy by having damage bubble up correctly.

Damage tracking has been implemented in a somewhat naïve fashion, but I believe it is a fair start.

Note that this currently depends on a fork of glutin where support has been added.

Fixes #3186.

Linux, X11. Using eb231b3e70b87875df4bdd1974d5e94704024d70, release build (the AUR one).

Starting up alacritty takes a significant amount of time. Simply doing time alacritty -e false reports about 390ms. Other terminal emulators, such as urxvt or termite start much faster, at about 160ms for both. It's also not just a matter of this simple benchmark, in normal usage it also feels slower to start.

Since you got so much pushback on the scrollback support, and mentioned possibly implementing it, I'm opening this as a tracking issue.

(ps: I also really want scrollback w/o tmux).

Using Alacritty (commit 6916537858d12e02478d618339d1b9f0e73a677a) on Ubuntu 16.04 with X11.

It is better described with the following screencast. Here I'm repeatedly opening and closing tig, so the output should be the same in each iteration. The screen often stays in a partially redrawed state (marked by mouse cursor movement). Then it finishes redraw correctly when I'm pressing a key.

System

OS: Arch Linux x86_64 Version: alacritty 0.11.0 (8dbaa0bb) Wayland, Sway 1.7

Details

This issue started recently, and as Sway version haven't changed since January 2022, I believe the issue is with recent updates of Alacritty.

Termdown is a countdown program, and recently I noticed that most of the time the screen stops being updated until I make any input (mouse movement or keyboard press). If I make the input, the terminal will update for a few seconds (or minutes if lucky) and stop again.

I debugged with alacritty -o debug.highlight_damage=true and recorded a video with the issue happening.

1. Run termdown -B 2m && notify-send 'Finish'
2. The terminal stopped being updated at 1m 58s.
3. After 2 minutes the notification is shown, but Alacritty screen is still stopped
4. I move my mouse cursor and Alacritty updates the screen

alacritty_update_bug.webm

This is the same change as https://github.com/kovidgoyal/kitty/commit/368bc91eed446daf9c32dfbce8aec718ebb1201d (also relevant: https://github.com/kovidgoyal/kitty/issues/4686) I confirmed that now the rendered color is exactly the same as Kitty.

I think it makes sense to make the default color space sRGB since other popular terminals are using sRGB by default including Kitty and iTerm. Also showing the same color as web browser, which is using sRGB, is beneficial for several apps such as nvim-colorizer.lua while doing web development.

Closes https://github.com/alacritty/alacritty/issues/4939

Verification on MacBook

I tested on MacBook Pro 14 inch, 2021 with pastel

• master (2bd26fbeb0c51cd8a98ae31d58ea105d1274387a)

  • We see that digital color meter (sRGB) is "out of range"
  • 

• this PR

  • We see the color is different from master and digital color meter (sRGB) is showing the "correct" value, which was the same value as the one shown on Kitty / iTerm (tested with the same pastel command) and Apple Safari (tested with <div style="background-color: #398810;">X</div>)
  • 

Concerns

Although I am personally happy with this patch since I am familiar with sRGB more than other color space (had no idea about P3 initially), this will be "breaking change" and some of macOS users will be unhappy due to changed colors for their colorscheme although changed one should be more consistent with other terminals. Actually it looks like Kitty made this configurable via macos_colorspace option so that people can still choose other color space such as P3 via configuration option.

Before merging this patch, we might need more discussions / considerations. (cf. https://github.com/kovidgoyal/kitty/issues/4686)

Linux: 6.0.8 kernel Fedora 38 rawhide Deskto: KDE Plasma (wayland)

#launched on nvidia 3070 via optimus env switch
#launching on igu amd 680m 6800hs has no issue
 __NV_PRIME_RENDER_OFFLOAD=1  /home/diego/.cargo/bin/alacritty -vv

#confirmed alacritty is bound to nvidia gpu
(base) [diego@MiWiFi-RB06-srv ~]$ nvidia-smi
Tue Jan  3 12:28:15 2023       
+-----------------------------------------------------------------------------+
| NVIDIA-SMI 525.60.11    Driver Version: 525.60.11    CUDA Version: 12.0     |
|-------------------------------+----------------------+----------------------+
| GPU  Name        Persistence-M| Bus-Id        Disp.A | Volatile Uncorr. ECC |
| Fan  Temp  Perf  Pwr:Usage/Cap|         Memory-Usage | GPU-Util  Compute M. |
|                               |                      |               MIG M. |
|===============================+======================+======================|
|   0  NVIDIA GeForce ...  Off  | 00000000:01:00.0 Off |                  N/A |
| N/A   48C    P8    12W /  N/A |      5MiB /  8192MiB |      0%      Default |
|                               |                      |                  N/A |
+-------------------------------+----------------------+----------------------+
                                                                               
+-----------------------------------------------------------------------------+
| Processes:                                                                  |
|  GPU   GI   CI        PID   Type   Process name                  GPU Memory |
|        ID   ID                                                   Usage      |
|=============================================================================|
|    0   N/A  N/A     46750      G   ...iego/.cargo/bin/alacritty        4MiB |

[0.000001743s] [INFO ] [alacritty] Welcome to Alacritty
[0.000067847s] [INFO ] [alacritty] Version 0.11.0
[0.000088776s] [INFO ] [alacritty] Running on Wayland
[0.000234178s] [INFO ] [alacritty_config_derive] No config file found; using default
[0.000570638s] [DEBUG] [alacritty] Loading "monospace" font
[0.007372359s] [DEBUG] [crossfont] Loaded Face Face { ft_face: Font Face: Regular, load_flags: TARGET_LIGHT, render_mode: "Lcd", lcd_filter: 1 }
[0.008381909s] [DEBUG] [crossfont] Loaded Face Face { ft_face: Font Face: Bold, load_flags: TARGET_LIGHT, render_mode: "Lcd", lcd_filter: 1 }
[0.009362214s] [DEBUG] [crossfont] Loaded Face Face { ft_face: Font Face: Regular, load_flags: NO_BITMAP | TARGET_LIGHT, render_mode: "Lcd", lcd_filter: 1 }
[0.010330316s] [DEBUG] [crossfont] Loaded Face Face { ft_face: Font Face: Bold, load_flags: NO_BITMAP | TARGET_LIGHT, render_mode: "Lcd", lcd_filter: 1 }
[0.011090949s] [DEBUG] [alacritty] Estimated scaling factor: 1
[0.011099205s] [DEBUG] [alacritty] Estimated window size: None
[0.011102321s] [DEBUG] [alacritty] Estimated cell size: 9 x 21
[0.011964615s] [DEBUG] [crossfont] Loaded Face Face { ft_face: Font Face: Regular, load_flags: TARGET_LIGHT, render_mode: "Lcd", lcd_filter: 1 }
[0.046511418s] [INFO ] [alacritty] Running on NVIDIA GeForce RTX 3070 Ti Laptop GPU/PCIe/SSE2
[0.046530384s] [INFO ] [alacritty] Using OpenGL ES 2.0 renderer
[0.046586318s] [INFO ] [alacritty] Using dual source blending
[0.047482075s] [INFO ] [alacritty] Display scale factor: 1
[0.047492324s] [DEBUG] [alacritty] Filling glyph cache with common glyphs
[0.049876026s] [INFO ] [alacritty] Cell size: 9 x 21
[0.049889441s] [INFO ] [alacritty] Padding: 0 x 0
[0.049892577s] [INFO ] [alacritty] Width: 800, Height: 600
[0.050028070s] [INFO ] [alacritty] PTY dimensions: 28 x 88
[0.051589483s] [INFO ] [alacritty] Initialisation complete
[destroyed object]: error 7: importing the supplied dmabufs failed
thread 'main' panicked at 'failed to dispatch wayland event queue: Os { code: 71, kind: Uncategorized, message: "Protocol error" }', /home/diego/.cargo/registry/src/github.com-1ecc6299db9ec823/alacritty-0.11.0/src/event.rs:1438:30
note: run with `RUST_BACKTRACE=1` environment variable to display a backtrace

System

OS: Ubuntu 20.04 Version: alacritty 0.11.0 (8dbaa0bb) Linux/BSD: gnome shell

I have been able to span alacritty terminal (v 0.10.1) over two monitors using the config file : window:

  # Window dimensions (changes require restart)
  dimensions:
    columns: 274
    lines: 126
                                                                                                                                                                                                                                                                              
  # Window position (changes require restart)
  position:
    x: 0
    y: 0

  # Window decorations
  decorations: none

This is not possible anymore with the newest release. Can this be fixed?

Logs

I'm quite honestly tired of looking at YAML files with broken indentation. While YAML is an awful format even for experienced users, clearly it's completely impossible for inexperienced ones to grasp its concept. And the documentation of our configuration file with everything commented out by default probably doesn't help.

Since as far as I know all previous issues with TOML have been fixed and escape sequences can be easily specified in TOML strings now, I think it would make sense to switch to TOML as our default format.

In a transitory period we could probably get away with just loading existing YAML files and writing them out as TOML, to allow for backward compatibility. This will probably produce somewhat messy TOML files, but I can't imagine an auto-generated TOML file would be much worse than a YAML file.

The only major drawbacks I see is that this is a rather big change that will no doubt annoy a bunch of people and some more advanced features of YAML (like anchors) are not supported in TOML. But nobody really needs anchors and I kinda like annoying people.

@kchibisov You proposed this a while ago on IRC, do you think this is a bad idea?

See https://github.com/alacritty/alacritty/issues/1217 for prior discussions.

Pressing Ctrl + / does nothing.

For whatever it's worth: Pressing Ctrl + v pastes whatever is in the clipboard.

So i can't bind C-_ for commenting/uncommenting in neovim.

System

OS: Windows 11 Version: alacritty 0.11.0 (d23330a) Shell: PowerShell 7.3.1

Logs

Crashes: STDERR, STDOUT Font/Terminal size: alacritty -vv Keyboard and bindings: alacritty --print-events

[37.859271800s] [INFO ] [alacritty] glutin event: MainEventsCleared
[37.859429400s] [INFO ] [alacritty] glutin event: RedrawEventsCleared
[37.978542200s] [INFO ] [alacritty] glutin event: NewEvents(WaitCancelled { start: Instant { t: 209698.4448825s }, requested_resume: None })
[37.978888300s] [INFO ] [alacritty] glutin event: WindowEvent { window_id: WindowId(WindowId(2099080)), event: KeyboardInput { device_id: DeviceId(DeviceId(0)), input: KeyboardInput { scancode: 53, state: Released, virtual_keycode: Some(Slash), modifiers: CTRL }, is_synthetic: false } }
[37.979271300s] [INFO ] [alacritty] glutin event: MainEventsCleared
[37.979413400s] [INFO ] [alacritty] glutin event: RedrawEventsCleared
[38.234473100s] [INFO ] [alacritty] glutin event: NewEvents(WaitCancelled { start: Instant { t: 209698.5649091s }, requested_resume: None })
[38.234748400s] [INFO ] [alacritty] glutin event: WindowEvent { window_id: WindowId(WindowId(2099080)), event: KeyboardInput { device_id: DeviceId(DeviceId(0)), input: KeyboardInput { scancode: 53, state: Pressed, virtual_keycode: Some(Slash), modifiers: CTRL }, is_synthetic: false } }
[38.235074700s] [INFO ] [alacritty] glutin event: MainEventsCleared
[38.235285500s] [INFO ] [alacritty] glutin event: RedrawEventsCleared
[38.347000700s] [INFO ] [alacritty] glutin event: NewEvents(WaitCancelled { start: Instant { t: 209698.8207637s }, requested_resume: None })
[38.347350500s] [INFO ] [alacritty] glutin event: WindowEvent { window_id: WindowId(WindowId(2099080)), event: KeyboardInput { device_id: DeviceId(DeviceId(0)), input: KeyboardInput { scancode: 53, state: Released, virtual_keycode: Some(Slash), modifiers: CTRL }, is_synthetic: false } }
[38.347761500s] [INFO ] [alacritty] glutin event: MainEventsCleared
[38.347943900s] [INFO ] [alacritty] glutin event: RedrawEventsCleared
[38.611652900s] [INFO ] [alacritty] glutin event: NewEvents(WaitCancelled { start: Instant { t: 209698.9334393s }, requested_resume: None })
[39.394845700s] [INFO ] [alacritty] glutin event: NewEvents(WaitCancelled { start: Instant { t: 209699.9011342s }, requested_resume: None })
[39.395189000s] [INFO ] [alacritty] glutin event: WindowEvent { window_id: WindowId(WindowId(2099080)), event: KeyboardInput { device_id: DeviceId(DeviceId(0)), input: KeyboardInput { scancode: 53, state: Released, virtual_keycode: Some(Slash), modifiers: CTRL }, is_synthetic: false } }
[39.395459400s] [INFO ] [alacritty] glutin event: MainEventsCleared
[39.395608200s] [INFO ] [alacritty] glutin event: RedrawEventsCleared
[39.409893600s] [INFO ] [alacritty] glutin event: NewEvents(WaitCancelled { start: Instant { t: 209699.9810826s }, requested_resume: None })
[39.410080900s] [INFO ] [alacritty] glutin event: WindowEvent { window_id: WindowId(WindowId(2099080)), event: ModifiersChanged((empty)) }
[39.410257600s] [INFO ] [alacritty] glutin event: WindowEvent { window_id: WindowId(WindowId(2099080)), event: KeyboardInput { device_id: DeviceId(DeviceId(0)), input: KeyboardInput { scancode: 29, state: Released, virtual_keycode: Some(LControl), modifiers: (empty) }, is_synthetic: false } }