Reverts two commits:
977cd530c7bb9551de93900170bdaec4601b1b5b
820b7e432b57cd08c49d2e76cce4cb78016f0418
These break build from source on Linux for two reasons:
1.) The systemd user service needs to be installed in the `share`
prefix, not the `lib` prefix. This lets it get picked up in `~/.local`
but is also correct for just standard FHS paths.
2.) The `ghostty` path in the systemd user service needs to be absolute.
We should interpolate in the build install prefix to form an absolute
path.
Reverts two commits:
977cd530c7bb9551de93900170bdaec4601b1b5b
820b7e432b57cd08c49d2e76cce4cb78016f0418
These break build from source on Linux for two reasons:
1.) The systemd user service needs to be installed in the `share`
prefix, not the `lib` prefix. This lets it get picked up in `~/.local`
but is also correct for just standard FHS paths.
2.) The `ghostty` path in the systemd user service needs to be absolute.
We should interpolate in the build install prefix to form an absolute
path.
There are two main improvements being made here. First, we move away
from using autohash and instead use a one-shot strategy similar to the
Style hashing. Since the GlyphKey includes the Metrics struct, which
contains quite a few fields, autohash was performing expensive and
unnecessary repeated updates.
The second improvement is actually just, not hashing Metrics. By
ignoring the Metrics field, we can fit the rest of the GlyphKey into a
64-bit packed struct and just return that as the hash! It ends up being
unique for each GlyphKey in renderGlyph, and is nearly a zero-cost
operation.
This ends up boosting the performance (on my machine at least), from
around 560fps to 590fps on the DOOM-fire benchmark.
Flatpak currently does not export systemd user units. As such, remove
references to it from D-Bus services to prevent D-Bus daemon from trying
to start a non-existent service.
Additionally, make sure that the D-Bus service name is correct for debug
builds.
Follow up to #7433
There are two main improvements being made here. First, we move away from using autohash and instead
use a one-shot strategy similar to the Style hashing. Since the GlyphKey includes the Metrics struct,
which contains quite a few fields, autohash was performing expensive and unnecessary repeated updates.
The second improvement is actually just, not hashing Metrics. By ignoring the Metrics field, we can
fit the rest of the GlyphKey into a 64-bit packed struct and just return that as the hash! It
ends up being unique for each GlyphKey in renderGlyph, and is nearly a zero-cost operation.
This ends up boosting the performance (on my machine at least), from around 560fps to 590fps on the
DOOM-fire benchmark.
Flatpak currently does not export systemd user units. As such, remove
references to it from D-Bus services to prevent D-Bus daemon from trying
to start a non-existent service.
Additionally, make sure that the D-Bus service name is correct for debug
builds.
Without waiting on the xdg-open process on linux/freebsd, we end up with
a defunct (zombie) process after each time we open a URL.
For example, after click on two URLs in a ghostty, here is the output of
`ps ux | grep xdg-open`:
```
pbui 8364 0.0 0.0 0 0 tty7 Z+ 05:03 0:00 [xdg-open] <defunct>
pbui 8453 0.0 0.0 0 0 tty7 Z+ 05:03 0:00 [xdg-open] <defunct>
```
Perhaps we should revisit 695bc30, which removed the wait in the first
place. On my machine running Alpine Linux 3.22, `xdg-open` does not stay
alive and finishes immediately, thus making it safe to call wait (and
not block). This is also the case on my other machine running Ubuntu
24.04: `xdg-open` launches the URL in a browser and terminates
immediately.
Either way, this process must be waited upon eventually. Otherwise, we
will accumulate a collection of defunct processes until the terminal
itself terminates.
Introduces host resources directory as a new concept: A directory
containing application resources that can only be accessed from the host
operating system. This is significant for sandboxed application runtimes
like Flatpak where shells spawned on the host should have access to
application resources to enable integrations.
Alongside this, apprt is now allowed to override the resources lookup
logic.
## Description
Yet another edge case in #2484
When macOS's "Private WiFi address" feature is enabled it'll change the
hostname to a mac address. Mac addresses look like URIs with a hostname
and port component, e.g. `12:34:56:78:90:12` where `:12` looks like port
`12`. However, mac addresses use hex numbers and as such can also
contain letters `a` through `f`. So, a mac address like
`ab💿ef🆎cd:ef` is valid, but will not be parsed as a URI, because
`:ef` is not a valid port.
This commit attempts to fix that by checking if the hostname is a valid
mac address when `std.Uri.parse()` fails and constructing a new
`std.Uri` struct using that information.
It's not perfect, but is equally compliant with the URI spec as
`std.Uri` currently is. Meaning not at all compliant 😅
## Testing instructions
### Unit tests
> [!IMPORTANT]
> I don't know if these tests are run in CI or if they're picked up by
`zig build test`. I get an unrelated crash that mentions `minidump` and
an invalid OSC command when I try to run `zig build test` on my mac.
1. Make sure `zig test src/os/hostname.zig` is passing.
### Manual testing instructions
#### Setup - Enable the "Private WiFi address" setting
> [!IMPORTANT]
> You must be connected to WiFi to be able to test this.
1. Open your mac's "System Settings".
2. Go to Network → Wi-Fi → Details.
<img width="710" alt="image"
src="https://github.com/user-attachments/assets/fe30cfe7-8e77-4421-8b36-2f7aab0918dd"
/>
3. Set the "Private Wi-Fi address" setting to `Rotating`.
<img width="710" alt="image"
src="https://github.com/user-attachments/assets/bd695c20-106c-46bd-8862-cbdce55fed6f"
/>
> [!IMPORTANT]
> Now you wait. The private Wi-Fi address will eventually rotate to a
mac address that ends with a non-digit, e.g. `0a`, `ff`, `e2`, etc.
You'll notice this when your shell integration stops working, e.g. you
open a new tab in Ghostty and the shell is in your home directory
instead of whichever directory you had open in your previous tab.
#### Testing the changes
1. Open Ghostty.
3. `cd` to any directory that isn't the default (usually `$HOME`)
directory, e.g. `cd Documents`.
4. Open a new tab (<kbd>Cmd+T</kbd>) or split (<kbd>Cmd+D</kbd>).
5. Assuming the setup steps have been followed you should:
* On `main`: land in `$HOME` in the new tab or split.
* On this branch: land in the same working directory as the original tab
or split.
Introduces host resources directory as a new concept: A directory
containing application resources that can only be accessed from the host
operating system. This is significant for sandboxed application runtimes
like Flatpak where shells spawned on the host should have access to
application resources to enable integrations.
Alongside this, apprt is now allowed to override the resources lookup
logic.
This adds a new CLI `ghostty +edit-config`. This will open the config
file in the user's specified `$EDITOR`. If Ghostty has never been
configured, this will also create the initial config file with some
default templated contents (the same as that which we introduced back in
Ghostty 1.0.1 or something -- not new behavior here).
This is useful on its own because it will find the correct configuration
path to open. If users are terminal users anyway (not a big stretch
since this is a terminal app), this will allow them to easily edit
config right away.
This is also forward looking: I want to replace our "Open Config" action
to open a Ghostty window executing this command so that users can edit
their config in a terminal editor. This has been heavily requested since
forever (short of a full GUI settings editor, which is not ready yet). I
don't do this in this PR but plan to in a future PR.
Even further forward looking: when we have an API, we can have
`edit-config` auto-reload the config on exit. This isn't possible today
but this is where we'd put that.
This adds a new CLI `ghostty +edit-config`. This will open the config
file in the user's specified `$EDITOR`. If Ghostty has never been
configured, this will also create the initial config file with some
default templated contents (the same as that which we introduced back in
Ghostty 1.0.1 or something -- not new behavior here).
This is useful on its own because it will find the correct configuration
path to open. If users are terminal users anyway (not a big stretch
since this is a terminal app), this will allow them to easily edit
config right away.
This is also forward looking: I want to replace our "Open Config" action
to open a Ghostty window executing this command so that users can edit
their config in a terminal editor. This has been heavily requested since
forever (short of a full GUI settings editor, which is not ready yet). I
don't do this in this PR but plan to in a future PR.
This PR is a collection of small improvements, cleanup, and fixes to the
renderer, as a follow-up to #7620.
### Summary of changes
- The nearly identical-between-backends `cell.zig` and `image.zig` are
now both unified.
- Shader pipeline preparation is now even DRYer in preparation for
future changes that will involve adding new pipelines, where I don't
want to be slowed down by having to add them in multiple places to get
things working. There is now a single source of truth at the top of
`shaders.zig` for what core pipelines are available and how they're
configured.
- Global background color drawn in a separate step before individual
cell background colors, this is required to be able to draw kitty images
between these layers. Can't use the clear color for this because that
would require color space conversions on the CPU-side which we don't yet
have utilities for.
- Fixed a bug with the kitty image z-index logic where not having
foreground images made the background images act as foreground images.
- Moved the custom shader uniform buffer to the frame state so we don't
create a new one every frame like we were before.
- Fixed color glyphs under OpenGL being channel-swapped because the
texture format was RGBA when the data was BGRA (#7670).
Forgot to change these instances when I renamed rgb(a) to bgr(a), which
was breaking test builds. Also went ahead and fixed some code that was
assuming rgba was actually rgba order and added a note to another part.
Also changes color atlas to always use an sRGB internal format so that
the texture reads automatically linearize the colors.
Renames the misleading `rgba` atlas format to `bgra`, since both
FreeType and CoreText are set up to draw color glyphs in bgra.
In this format it will be a lot easier to iterate on this since adding
and removing pipelines only has to be done in a single place.
This commit also separates out the main background color from individual
cell background color drawing, because sometimes kitty images need to be
between the main background and individual cell backgrounds (kitty image
z-index seems to be entirely broken at the moment, I intend to fix it in
future commits).
The code in metal/image.zig and opengl/image.zig was virtually identical
save for the texture options, so I've moved that to the GraphicsAPI and
unified them in to renderer/image.zig
Fixes#7647
See #7647 for context. This commit works by extending the `input` work
introduced in #7652 to libghostty so that the macOS can take advantage
of it. At that point, it's just the macOS utilizing `input` in order to
set the command and `exit` up similar to Terminal and iTerm2.
This applies both to files opened directly by Ghostty as well as the App
Intent to run a command in a new terminal.
Fixes#7647
See #7647 for context. This commit works by extending the `input` work
introduced in #7652 to libghostty so that the macOS can take advantage
of it. At that point, its just the macOS utilizing `input` in order to
set the command and `exit` up similar to Terminal and iTerm2.
This adds a new configuration `input` that allows passing either raw
text or file contents as stdin when starting the terminal.
The input is sent byte-for-byte to the terminal, so control characters
such as `\n` will be interpreted by the shell and can be used to run
programs in the context of the loaded shell. After the input is sent, it
switches to `stdin` as usual.
Example: `ghostty --input="hello, world\n"` will start the your default
shell, run `echo hello, world`, and then show the prompt
The `--input` configuration can be repeated to send multiple data
concatenated directly to together. Values can be also be prefixed with
`file:` to send a file contents, e.g. `file:data.txt`. File contents are
limited to 10MB to prevent excessive memory usage and avoid malicious
use. Beyond that, users should write their own script...
This is all in pursuit of #7647, but we're not there yet.
This adds a new configuration `input` that allows passing either raw
text or file contents as stdin when starting the terminal.
The input is sent byte-for-byte to the terminal, so control characters
such as `\n` will be interpreted by the shell and can be used to run
programs in the context of the loaded shell.
Example: `ghostty --input="hello, world\n"` will start the your default
shell, run `echo hello, world`, and then show the prompt.
Supersedes #6912, implements #6901
Also included in this PR is a fix/cleanup of the custom shader uniform
handling, moved the CPU-side custom shader uniforms struct to the main
renderer struct instead of having it be per-frame, moved the layout
struct to `shadertoy.zig` since it has the `std140` layout for both
backends.
Also, I added the current/previous cursor colors to the uniforms, since
I figured they'd be useful to have and it's a trivial addition.
### Future Work
- This extension to the shadertoy uniforms needs to be documented
somewhere so it's discoverable by users.
- The flipped Y axis on Metal still needs to be fully addressed instead
of just being patched over like it is right now.
This regression may have existed on Sequoia too, but I only saw it on
Tahoe.
The issue is that we should not be setting up titlebar accessory views
when there is no titlebar. This was triggering an AppKit assertion.
To further simplify things, I always use the basic window styling if
window decorations are off, too.
This regression may have existed on Sequoia too, but I only saw it on
Tahoe.
The issue is that we should not be setting up titlebar accessory views
when there is no titlebar. This was triggering an AppKit assertion.
To further simplify things, I always use the basic window styling if
window decorations are off, too.
These should not be independent per-frame, that makes the time
calculations all sorts of jank.
Also moves the uniforms struct layout in to `shadertoy.zig` and cleans
up the handling in general somewhat.
It's here, the long-foretold and long-procrastinated renderer rework!
Hopefully this makes it easier to adapt and modify the renderer in the
future and ensures feature parity between Metal and OpenGL. Despite
having been a lot of work to write initially, with the abstraction layer
in place I feel like working on the renderer will be a much more
pleasant experience going forward.
## Key points
- CPU-side renderer logic is now mostly unified via a generic
`Renderer`.
- A graphics API abstraction layer over OpenGL and Metal has been
introduced.
- Minimum OpenGL version bumped to `4.3`, so can no longer be run on
macOS; I used the nix VM stuff for my testing during development. (Edit
by @mitchellh: Note for readers that Ghostty still works on macOS, but
the OpenGL backend doesn't, only the Metal one)
- The OpenGL backend now supports linear blending! Woohoo! The default
`alpha-blending` has been updated to `linear-corrected` since it's
essentially a strict improvement over `native`. The default on macOS is
still `native` though to match other mac apps in appearance, since macOS
users are more sensitive to text appearance.
- Custom shaders can now be hot reloaded.
- The background color is once again drawn by us, so custom shaders can
interact with it properly. In general, custom shaders should be a little
more robust.
## The abstraction layer
The general hierarchy of the abstraction layer is as such:
```
[ GraphicsAPI ] - Responsible for configuring the runtime surface
| | and providing render `Target`s that draw to it,
| | as well as `Frame`s and `Pipeline`s.
| V
| [ Target ] - Represents an abstract target for rendering, which
| could be a surface directly but is also used as an
| abstraction for off-screen frame buffers.
V
[ Frame ] - Represents the context for drawing a given frame,
| provides `RenderPass`es for issuing draw commands
| to, and reports the frame health when complete.
V
[ RenderPass ] - Represents a render pass in a frame, consisting of
: one or more `Step`s applied to the same target(s),
[ Step ] - - - - each describing the input buffers and textures and
: the vertex/fragment functions and geometry to use.
:_ _ _ _ _ _ _ _ _ _/
v
[ Pipeline ] - Describes a vertex and fragment function to be used
for a `Step`; the `GraphicsAPI` is responsible for
these and they should be constructed and cached
ahead of time.
[ Buffer ] - An abstraction over a GPU buffer.
[ Texture ] - An abstraction over a GPU texture.
```
More specific documentation can be found on the relevant structures.
## Miscellany
Renderers (which effectively just means the generic renderer) are now
expected to only touch GPU resources in `init`, certain lifecycle
functions such as the `displayRealized`/`displayUnrealized` callbacks
from GTK-- and `drawFrame`; and are also expected to be thread-safe.
This allows the renderer thread to build the CPU-side buffers
(`updateFrame`) even if we can only *draw* from the app thread.
Because of this change, we can draw synchronously from the main thread
on macOS when necessary to always have a frame of the correct size
during a resize animation. This was necessary to allow the background to
be drawn by our GPU code (instead of setting a background color on the
layer) while still avoiding holes during resize.
The OpenGL backend now theoretically has access to multi-buffering, but
it's disabled (by setting the buffer count to 1) because it
synchronously waits for frames to complete anyway which means that the
extra buffers were just a waste of memory.
## Validation
To validate that there are no significant or obvious problems, I
exercised both backends with a variety of configurations, and visually
inspected the results. Everything looks to be in order.
The images are available in a gist here:
https://gist.github.com/qwerasd205/c1bd3e4c694d888e41612e53c0560179
## Memory
Here's a comparison of memory usage for ReleaseFast builds on macOS,
between `main` and this branch.
Memory figures given are values from Activity Monitor measuring windows
of the same size, with two tabs with 3 splits each.
||Before|After|
|-:|-|-|
|**Memory**|247.9 MB|224.2 MB|
|**Real Memory**|174.4 MB|172.5 MB|
Happily, the rework has slightly *reduced* the memory footprint- likely
due to removing the overhead of `CAMetalLayer`. (The footprint could be
reduced much further if we got rid of multi-buffering and satisfied
ourselves with blocking for each frame, but that's a discussion for
another day.)
If someone could do a similar comparison for Linux, that'd be much
appreciated!
## Notes / future work
- There are a couple structures that *can* be unified using the
abstraction layer, but I haven't gotten around to unifying yet.
Specifically, in `renderer/(opengl|metal)/`, there's `cell.zig` and
`image.zig`, both of which are substantially identical between the two
backends. `shaders.zig` may also be a candidate for unification, but
that might be *overly* DRY.
- ~~I did not double-check the documentation for config options, which
may mention whether certain options can be hot-reloaded; if it does then
that will need to be updated.~~ Fixed: be5908f
- The `fragCoord` for custom shaders originates at the top left for
Metal, but *bottom* left for OpenGL; fixing this will be a bit annoying,
since the screen texture is likewise vertically flipped between the two.
Some shaders rely on the fragcoord for things like falling particles, so
this does need to be fixed.
- `Target` should be improved to support multiple types of targets right
now it only represents a framebuffer or iosurface, but it should also be
able to represent a texture; right now a kind of messy tagged union is
used so that steps can accept both.
- Custom shader cursor uniforms (#6912) and terminal background images
(#4226, #5233) should be much more straightforward to implement on top
of this rework, and I plan to make follow-up PRs for them once this is
merged.
- I *do* want to do a rework of the pipelines themselves, since the way
we're rendering stuff is a bit messy currently, but this is already a
huge enough PR as it is- so for now the renderer still uses the same
rendering passes that Metal did before.
- We should probably add a system requirements section to the README
where we can note the minimum required OpenGL version of `4.3`, any even
slightly modern Linux system will support this, but it would be good to
document it somewhere user-facing anyway.
# TODO BEFORE MERGE
- [x] Have multiple people test this on both macOS and linux.
- [ ] ~~Have someone with a better dev setup on linux check for memory
leaks and other problems.~~ (Skipped, will merge and let tip users
figure this out, someone should *specifically* look for memory leaks
before the next versioned release though.)
- [x] Address any code review feedback.
