This is a big'un.
- **Glyph constraint logic is now done fully on the CPU** at the
rasterization stage, so it only needs to be done once per glyph instead
of every frame. This also lets us eliminate padding between glyphs on
the atlas because we're doing nearest-neighbor sampling instead of
interpolating-- which ever so slightly increases our packing efficiency.
- **Special constraints for nerd font glyphs** are applied based roughly
on the constraints they use in their patcher. It's a simplification of
what they do, the largest difference being that they scale groups of
glyphs based on a shared bounding box so that they maintain relative
size to one another, but that would require loading all glyphs on the
group and I'd want to do that on font load TBH and at that point I'd
basically be re-implementing the nerd fonts patcher in Zig to patch
fonts at load time which is way beyond the scope I want to have. (Fixes
#7069)
- These constraints allow for **perfectly sized and centered emojis**,
this is very nice.
- **Changed the default embedded fonts** from 4 copies (regular, italic,
bold, bold italic) of a patched (and outdated) JetBrains Mono to a
single JetBrains Mono variable font and a single Nerd Fonts Symbols Only
font. This cuts the weight of those down from 9MB to 3MB!
- **FreeType's `renderGlyph` is significantly reworked**, and the new
code is, IMO, much cleaner- although there are probably some edge case
behavior differences I've introduced.
> [!NOTE]
> One breaking change I definitely introduced is changing the
`monochrome` freetype load flag config from its previous completely
backwards meaning to instead the correct one (I also changed the
default, so this won't affect any user who hasn't touched it, but users
who set the `monochrome` flag will find their fonts quite crispy after
this change because they will have no anti-aliasing anymore)
### Future work
Following this change I want to get to work on automatic font size
matching (a la CSS
[`font-size-adjust`](https://developer.mozilla.org/en-US/docs/Web/CSS/font-size-adjust)).
I set the stage for that quite some time ago so it shouldn't be too much
work and it will be a big benefit for users who regularly use multiple
writing systems and so have multiple fonts for them that aren't
necessarily size-compatible.
This deletes the GLFW apprt from the Ghostty codebase.
The GLFW apprt was the original apprt used by Ghostty (well, before
Ghostty even had the concept of an "apprt" -- it was all just a single
application then). It let me iterate on the core terminal features,
rendering, etc. without bothering about the UI. It was a good way to get
started. But it has long since outlived its usefulness.
We've had a stable GTK apprt for Linux (and Windows via WSL) and a
native macOS app via libghostty for awhile now. The GLFW apprt only
remained within the tree for a few reasons:
1. Primarily, it provided a faster feedback loop on macOS because
building the macOS app historically required us to hop out of the
zig build system and into Xcode, which is slow and cumbersome.
2. It was a convenient way to narrow whether a bug was in the
core Ghostty codebase or in the apprt itself. If a bug was in both
the glfw and macOS app then it was likely in the core.
3. It provided us a way on macOS to test OpenGL.
All of these reasons are no longer valid. Respectively:
1. Our Zig build scripts now execute the `xcodebuild` CLI directly and
can open the resulting app, stream logs, etc. This is the same
experience we have on Linux. (Xcode has always been a dependency of
building on macOS in general, so this is not cumbersome.)
2. We have a healthy group of maintainers, many of which have access
to both macOS and Linux, so we can quickly narrow down bugs
regardless of the apprt.
3. Our OpenGL renderer hasn't been compatible with macOS for some time
now, so this is no longer a useful feature.
At this point, the GLFW apprt is just a burden. It adds complexity
across the board, and some people try to run Ghostty with it in the real
world and get confused when it doesn't work (it's always been lacking in
features and buggy compared to the other apprts).
So, it's time to say goodbye. Its bittersweet because it is a big part
of Ghostty's history, but we've grown up now and it's time to move on.
Thank you, goodbye.
(NOTE: If you are a user of the GLFW apprt, then please fork the project
prior to this commit or start a new project based on it. We've warned
against using it for a very, very long time now.)
Rather than using binaries statically in our source tree; this makes
them easier to update. This also makes it so that they are separated
from each other rather than using a patched JB mono as our fallback.
This update also fixes a memory leak that was caused by blocks not being
deallocated and just collecting every single frame, slowly accumulating
memory until OOM.
We need to use this version of z2d so that we can get reproducible PNG
exports in CI for testing, since previously the PNG export was affected
by the CPU arch / features because it depended on vector width.
Fixes#7724
Background at the end of the commit message. The fix in libxev is
described in the PR and commit we pin to here, but basically we swap
read for poll for eventfd/timerfd.
From Jens Axboe on X:
> This will fix it: https://pastebin.com/n7JSZWpW which makes me suspicious
> that it's an S_IFREG check somewhere else, as anon inodes are now listed as
> regular files. Has potentially pretty broad implications...
> I think I can already answer why that breaks things - io_uring checks if
> this is a regular file, and if it is, it doesn't do short reads. Short
> reads on regular files (or a bdev) will cause application issues, as
> basically nobody expects them.
> Now we have what acts like a char dev, but where io_uring will retry IO
> because the application asked for more data than what was delivered. This
> will cause the weird slowdowns as data isn't delivered as soon as it's
> available.
Lazy dependencies are only fetched if the build script would actually
reach a usage of that dependency at runtime (when the `lazyDependency`
function is called). This can save a lot of network traffic, disk uage,
and time because we don't have to fetch and build dependencies that we
don't actually need.
Prior to this commit, Ghostty fetched almost everything for all
platforms and configurations all the time. This commit reverses that to
fetching almost nothing until it's actually needed.
There are very little downsides to doing this[1]. One downside is `zig
build --fetch` doesn't fetch lazy dependencies, but we don't rely on
this command for packaging and suggest using our custom shell script
that downloads a cached list of URLs (`build.zig.zon.txt`).
This commit doesn't cover 100% of dependencies, since some provide no
benefit to make lazy while the complexity to make them lazy is higher
(in code style typically).
Conversely, some simple dependencies are marked lazy even if they're
almost always needed if they don't introduce any real complexity to the
code, because there is very little downside to do so.
[1]: https://ziggit.dev/t/lazy-dependencies-best-dependencies/5509/5
Closes#6702
This removes our mach-glfw dependency and replaces it with an in-tree
pkg/glfw that includes both the source for compiling glfw as well as the
Zig bindings. This matches the pattern from our other packages.
This is based on the upstream mach-glfw work and therefore includes the
original license and copyright information.
The reasoning is stated in the issue but to summarize for the commit:
- mach-glfw is no longer maintained, so we have to take ownership
- mach-glfw depended on some large blobs of header files to enable
cross-compilation but this isn't something we actually care about,
so we can (and do) drop the blobs
- mach-glfw blobs were hosted on mach hosts. given mach-glfw is
unmaintained, we can't rely on this hosting
- mach-glfw relied on a "glfw" package which was owned by another
person to be Zig 0.14 compatible, but we no longer need to rely on
this
- mach-glfw builds were outdated based on latest Zig practices
As of now `gtk4-layer-shell` is unavailable on recent, stable releases
of many distros (Debian 12, Ubuntu 24.04, openSUSE Leap & Tumbleweed, etc.)
and outdated on many others (Nixpkgs 24.11/unstable, Fedora 41, etc.)
This is inconvenient for our users and severely limits where the quick
terminal can be used. As a result we then build gtk4-layer-shell ourselves
by default unless `--system` or `-fsys=gtk4-layer-shell` are specified.
This also allows me to add an idiomatic Zig API on top of the library
and avoiding adding even more raw C code in the GTK apprt.
Since we now build gtk4-layer-shell it should be theoretically available
on all Linux systems we target. As such, the `-Dgtk-layer-shell` build
option has been removed. This is somewhat of an experimental change as
I don't know if gtk4-layer-shell works perfectly across all distros, and
we can always add the option back if need be.
Related to #3224
Previously, Ghostty used a static API for async event handling: io_uring
on Linux, kqueue on macOS. This commit changes the backend to be dynamic
on Linux so that epoll will be used if io_uring isn't available, or if
the user explicitly chooses it.
This introduces a new config `async-backend` (default "auto") which can
be set by the user to change the async backend in use. This is a
best-effort setting: if the user requests io_uring but it isn't
available, Ghostty will fall back to something that is and that choice
is up to us.
Basic benchmarking both in libxev and Ghostty (vtebench) show no
noticeable performance differences introducing the dynamic API, nor
choosing epoll over io_uring.
Add a `+boo` command to show the animation from the website. The data
for the frames is compressed during the build process. This build step
was added to the SharedDeps object because it is used in both
libghostty and in binaries.
The compression is done as follows:
- All files are concatenated together using \x01 as a combining byte
- The files are compressed to a cached build file
- A zig file is written to stdout which `@embedFile`s the compressed
file and exposes it to the importer
- A new anonymous module "framedata" is added in the SharedDeps object
Any file can import framedata and access the compressed bytes via
`framedata.compressed`. In the `boo` command, we decompress the file and
split it into frames for use in the animation.
The overall addition to the binary size is 348k.
This adds a new script `update-mirror.sh` which generates the proper
blob format for R2 (or any blob storage) to mirror all of our
dependencies.
It doesn't automate updating build.zig.zon but on an ongoing basis this
should be easy to do manually, and we can strive to automate it in the
future.
I omitted iTerm2 color themes because we auto-update that via CI and
updating all of the machinery to send it to our mirror and so on is a
pain. Additionally, this doesn't mirror transitive dependencies because
Zig doesn't have a way to fetch those from a mirror instead (unless you
pre-generate a full cache like packagers but that's not practical for
day to day development).
It's hugely beneficial just to get most of our dependencies mirrored.
This fixes a regression in 1.1.1/1.1.2 where our PACKAGING docs mention
using `fetch-zig-cache.sh` but it was removed. This commit adds it back,
generating its contents from the build.zig.zon file (via zon2nix which
we use for our Nix packaging).
For packagers, there are no dependency changes: you still need Zig and
POSIX sh. For release time, Ghostty has a new dependency on `jq` but
otherwise the release process is the same. The check-zig-cache.sh script
is updated to generate the new build.zig.zon.txt file.
