This makes it so `zig build run` can take arguments such as
`--config-default-files=false` or any other configuration. Previously,
it only accepted commands such as `+version`.
Incidentally, this also makes it so that the app in general can now take
configuration arguments via the CLI if it is launched as a new instance
via `open`. For example:
open -n Ghostty.app --args --config-default-files=false
This previously didn't work. This is kind of cool.
To make this work, the libghostty C API was modified so that
initialization requires the CLI args, and there is a new C API to try to
execute an action if it was set.
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.
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.
This commit is very large, representing about a month of work with many
interdependent changes that don't separate cleanly in to atomic commits.
The main change here is unifying the renderer logic to a single generic
renderer, implemented on top of an abstraction layer over OpenGL/Metal.
I'll write a more complete summary of the changes in the description of
the PR.
This is recommended for macOS Tahoe and all standard menu items now have
associated images. This makes our app look more polished and native for
macOS Tahoe.
For icon choice, I tried to copy other native macOS apps as much as
possible, mostly from Xcode. It looks like a lot of apps aren't updated
yet. I'm absolutely open to suggestions for better icons but I think
these are a good starting point.
One menu change is I moved "reset font size" above "increase font size"
which better matches other apps (e.g. Terminal.app).
I've only recently been using programs that use user notifications heavily
and this commit addresses a number of annoyances I've encountered.
1. Notifications dispatched while the source terminal surface is
focused are now only shown for a short time (3 seconds hardcoded)
and then automatically dismiss.
2. Notifications are dismissed when the target surface becomes focused
from an unfocused state. This dismissal happens immediately (no
delay).
3. Notifications are dismissed when the application exits.
4. This fixes a bug where notification callbacks were modifying view
state, but the notification center doesn't guarantee that the
callback is called on the main thread. We now ensure that
the callback is always called on the main thread.
This is a major rework of how we represent, handle, and render splits in
the macOS app.
This new PR moves the split structure into a dedicated, generic
(non-Ghostty-specific) value-type called `SplitTree<V>`. All logic
associated with splits (new split, close split, move split, etc.) is now
handled by notifications on `BaseTerminalController`. The view hierarchy
is still SwiftUI but it has no logic associated with it anymore and
purely renders a static tree of splits.
Previously, the split hierarchy was owned by AppKit in a type called
`SplitNode` (a recursive class that contained the tree structure). All
logic around creating, zooming, etc. splits was handled by notification
listeners directly within the SwiftUI hierarchy. SwiftUI managed a
significant amount of state and we heavily used bindings, publishers,
and more. The reasoning for this is mostly historical: splits date back
to when Ghostty tried to go all-in on SwiftUI. Since then, we've taken a
more balanced approach of SwiftUI for views and AppKit for data and
business logic, and this has proven a lot more maintainable.
## Spatial Navigation
Previously, focus moving was handled by traversing the tree structure.
This led to some awkward behaviors. See:
https://github.com/ghostty-org/ghostty/issues/524#issuecomment-2668396095
In this PR, we now handle focus moving spatially. This means that move
"left" means moving to the visually left split (from the top-left
corner, a future improvement would be to do it from the cursor
position).
Concretely, given the following split structure:
```
+----------+-----+
| | b |
| | |
| a +-----+
| | |
| | |
| | |
| | |
|----------| d |
| c | |
| | |
+----------+-----+
```
Moving "right" from `c` now moves to `d`. Previously, it would go to
`b`. On Linux, it still goes to `b`.
## Value Types
One of the major architectural shifts is moving **purely to immutable
value types.** Whenever a split property changes such as a new split,
the ratio between splits, zoomed state, etc. we _create an entirely new
`SplitTree` value_ and replace it along the entire view hierarchy. This
is in some ways wasteful, but split hierarchies are relatively small
(even the largest I've seen in practical use are dozens of splits, which
is small for a computer). And using value types lets us get rid of a ton
of change notification soup around the SwiftUI hierarchy. We can rely on
reference counting to properly clean up our closed views.
> [!NOTE]
>
> As an aside, I think value types are going to make it a lot easier in
the future to implement features like "undo close." We can just keep a
trailing list of surface tree states and just restore them. This PR
doesn't do anything like that, but it's now possible.
## SwiftUI Simplicity
Our SwiftUI view hierarchy is dramatically simplified. See the
difference in `TerminalSplitTreeView` (new) vs `TerminalSplit` (old).
There's so much less logic in our new views (almost none!). All of it is
in the AppKit layer which is just way nicer.
## AI Notes
This PR was heavily written by AI. I reviewed every line of code that
was rewritten, and I did manually rewrite at every step of the way in
minor ways. But it was very much written in concert. Each commit usually
started as an AI agent writing the whole commit, then nudging to get
cleaned up in the right way.
One thing I found in this task was that until the last commit, I kept
the entire previous implementation around and compiling. The agent
having access to a previous working version of code during a refactor
made the code it produced as follow up in the new architecture
significantly better, despite the new architecture having major
fundamental differences in how it works!
Conversion of #7497 to a PR. This implements a feature requested in
#7331: an option to hide the default window buttons on macOS for a
cleaner aesthetic.
~~Builds on #7502 as it requires the same change to avoid the main
toolbar title showing on top of the tab bar.~~ EDIT: rebased on main now
that #7502 was merged.
I aligned the scope of the new option with `macos-titlebar-style`, since
they both customize titlebar elements. This means it has the same edge
case quirks: For example, if you change the setting, reload the config,
and then open a new tab, the appearance of the current window will
depend on which tab is in the foreground. I did it this way because
`macos-titlebar-style` provided an easy template for which derived
configs and functions to modify. Let me know if you want me to try
adjusting this so that a change in the setting also takes effect for
current windows/tabs, which I _think_ should be possible.
Screenshots:
* `macos-titlebar-style = transparent` (default)
* `macos-titlebar-style = tabs`
This fixes a small memory leak I found where the `SplitNode.Leaf` was
not being deinitialized properly when closing a split. It would get
deinitialized the next time a split was made or the window was closed,
so the leak wasn't big. The surface view underneath the split was also
properly deinitialized because we forced it, so again, the leak was
quite small.
But conceptually this is a big problem, because when we change the
surface tree we expect the deinit chain to propagate properly through
the whole thing, _including_ to the SurfaceView.
This fixes that by removing the `id(node)` call. I don't find this to be
necessary anymore. I don't know when that happened but we've changed
quite a lot in our split system since it was introduced. I'm also not
100% sure why the `id(node)` was causing a strong reference to begin
with... which bothers me a bit.
AI note: While I manually hunted this down, I started up Claude Code and
Codex in separate tabs to also hunt for the memory leak. They both
failed to find it and offered solutions that didn't work.
Fixes#7337
AppKit encodes functional keys as PUA codepoints. We don't want to send
that down as valid text encoding for a key event because KKP uses that
in particular to change the encoding with associated text.
I think there may be a more specific solution to this by only doing this
within the KKP encoding part of KeyEncoder but that was filled with edge
cases and I didn't want to risk breaking anything else.
