We previously used a readonly variable (__ghostty_ps0) to define the
best __ghostty_preexec_hook expansion for the current bash version.
This works pretty well, but it had the downside of managing another
variable (#11258).
We can instead simplify this a bit by moving this into __ghostty_hook. I
didn't take that approach originally because I wanted to avoid the bash
version check on each command, but slightly loosening our guard check to
just look for "__ghostty_preexec_hook" (rather than the full expansion
expression) means we can bury the bash version check to the cold path.
One small gap here is that we may not update PS0 to the correct syntax
if we start switching between significantly different bash versions in
interactive subshells, but that seems like a pretty rare case to handle
given the benefits of this approach.
ble.sh performs its own cursor positioning so we get multiple newlines
with 133;A's fresh-line behavior. ble.sh is a large enough project to
justify this additional, unambiguous conditional.
See: akinomyoga/ble.sh#684
See: wezterm/wezterm#5072
ble.sh performs its own cursor positioning so we get multiple newlines
with 133;A's fresh-line behavior. ble.sh is a large enough project to
justify this additional, unambiguous conditional.
See: akinomyoga/ble.sh#684
See: wezterm/wezterm#5072
Passing a `token` value causes this action to use the GitHub REST API,
which is subject to rate limits. We can chew through that allowance
quickly (1,000 requests/hour) given that we run two of these actions per
workflow run.
`token` defaults to the workflow's token, but by setting it explicitly
to an empty string, the action will instead use `git diff` to determine
the modified paths. This works fine for our case because we're already
running the checkout action, so we have an up-to-date repository view.
This also has the advantage of working around the 300 files GitHub REST
API limit for listing changed files.
Ref: https://github.com/dorny/paths-filter
Passing a `token` value causes this action to use the GitHub REST API,
which is subject to rate limits. We can chew through that allowance
quickly (1,000 requests/hour) given that we run two of these actions per
workflow run.
`token` defaults to the workflow's token, but by setting it explicitly
to an empty string, the action will instead use `git diff` to determine
the modified paths. This works fine for our case because we're already
running the checkout action, so we have an up-to-date repository view.
This also has the advantage of working around the 300 files GitHub REST
API limit for listing changed files.
Ref: https://github.com/dorny/paths-filter
We need to handle on more case: when an existing PROMPT_COMMAND ends in
a newline, we don't want to append a ; because that already counts as a
command separator.
We now handle all of these PROMPT_COMMAND cases:
- Ends with ; — no ; added
- Ends with \n or other whitespace — no ; added
- Ends with a command name — ; added as separator
See: #11245
This moves all our examples away from embedded source to `@snippet` and
files so that we can use our CI to actually run the builds and keep them
working.
Note: I used AI to extract the examples, and it did some weird merging
stuff. It all works but I want to make sure all these examples are still
human friendly so I need to go back and review all that. I clicked
through the web docs and they look good, just need to verify the GitHub
flow.
We need to handle on more case: when an existing PROMPT_COMMAND ends in
a newline, we don't want to append a ; because that already counts as a
command separator.
We now handle all of these PROMPT_COMMAND cases:
- Ends with ; — no ; added
- Ends with \n or other whitespace — no ; added
- Ends with a command name — ; added as separator
See: #11245
The dynamic example directory discovery added in bb3b3ba included
all subdirectories under example/, but some (wasm-key-encode,
wasm-sgr) are pure HTML examples with no build.zig.zon. Running
zig build in those directories falls back to the root build.zig
and attempts a full GTK binary build, which fails on CI.
Filter the listing to only include directories that contain a
build.zig.zon file so non-Zig examples are excluded from the
build matrix.
Extract inline @code blocks from vt headers (size_report.h, modes.h,
sgr.h, paste.h, mouse.h, key.h) into standalone buildable examples
under example/. Each header now uses Doxygen @snippet tags to include
code from the example source files, keeping documentation in sync
with code that is verified to compile and run.
New example projects: c-vt-size-report and c-vt-modes. Existing
examples (c-vt-sgr, c-vt-paste, c-vt-mouse-encode, c-vt-key-encode)
gain snippet markers so their code can be referenced from the headers.
Conceptual snippets in key.h, mouse.h, and key/encoder.h that show
terminal-state usage patterns remain inline since they cannot be
compiled standalone.
Replace the hardcoded matrix list in the build-examples job with a
dynamic list-examples job that discovers all subdirectories under
example/ at runtime. This uses ls/jq to produce a JSON array and
fromJSON() to feed it into the matrix, so new examples are picked
up automatically without updating the workflow.
Extract the inline code example from focus.h into a standalone
buildable example at example/c-vt-encode-focus. The header now
uses a Doxygen @snippet tag to include the code from the example
source file, so the documentation stays in sync with code that
is verified to compile and run.
Extract size report encoding into a reusable module and expose it
through the libghostty-vt C API as `ghostty_size_report_encode()`.
Size report escape sequences (mode 2048 in-band reports, XTWINOPS CSI
14/16/18 t responses) were formatted inline in
`Termio.sizeReportLocked`, and `termio.Message` carried its own
duplicate enum for report styles. This made the encoding logic
impossible to reuse from the C library and kept the style type
unnecessarily scoped to termio.
## Example
```c
GhosttySizeReportSize size = {
.rows = 24, .columns = 80,
.cell_width = 9, .cell_height = 18,
};
char buf[64];
size_t written = 0;
ghostty_size_report_encode(
GHOSTTY_SIZE_REPORT_MODE_2048, size,
buf, sizeof(buf), &written);
// buf contains: "\x1b[48;24;80;432;720t"
```
Add ghostty_size_report_encode() to libghostty-vt, following the
same pattern as focus encoding: a single stateless function that
writes a terminal size report escape sequence into a caller-provided
buffer.
The size_report.zig Style enum and Size struct now use lib.Enum and
lib.Struct so the types are automatically C-compatible when building
with c_abi, eliminating the need for duplicate type definitions in
the C wrapper. The C wrapper in c/size_report.zig re-exports these
types directly and provides the callconv(.c) encode entry point.
Supports mode 2048 in-band reports and XTWINOPS responses (CSI 14 t,
CSI 16 t, CSI 18 t).
Size report escape sequences were previously formatted inline in
Termio.sizeReportLocked, and termio.Message carried a duplicate enum for
report styles. That made the encoding logic harder to reuse and kept
the style type scoped to termio.
Move the encoding into terminal.size_report and export it through
terminal.main. The encoder now takes renderer.Size directly and derives
grid and pixel dimensions from one source of truth. termio.Message now
aliases terminal.size_report.Style, and Termio writes reports via the
shared encoder.
Fixes [#11935.](https://github.com/ghostty-org/ghostty/issues/11395)
I’m new to Zig, so I used AI assistance (Codex) while preparing this
change. Before opening this PR, I manually reviewed every line of the
final patch and stepped through the parser in LLDB to verify the
behavior. Happy to make any changes.
To better understand the parser, I also built a small model-checker
model
[here](https://gist.github.com/wyounas/284036272ba5893b6e413cafe2fe2a24).
Separately from this fix, I think formal verification and modeling could
be useful for parser work in Ghostty. The model is written in FizzBee,
which uses a Python-like Starlark syntax and is fairly readable. If that
seems useful, I’d be happy to open a separate discussion about whether
something like that belongs in the repository as executable
documentation or an additional safety net for future parser changes.
This is consistent with our bash prompt handling and also lets us
simplify our multiline prompt logic (because it no longer needs to work
around 133;A's fresh-line behavior).
This is consistent with our bash prompt handling and also lets us
simplify our multiline prompt logic (because it no longer needs to work
around 133;A's fresh-line behavior).
Previously libghostty-vt had no way for C consumers to query, set, or
report on terminal modes. Callers that needed to respond to DECRPM
requests or inspect mode state had no public interface to do so.
This adds three layers of mode support to the C API:
- `GhosttyMode` — a 16-bit packed type with inline helpers to construct
and inspect mode tags, plus `GHOSTTY_MODE_*` macros for all supported
ANSI and DEC private modes.
- `ghostty_terminal_mode_get` / `ghostty_terminal_mode_set` — query and
set mode values on a terminal handle.
- `ghostty_mode_report_encode` — encode a DECRPM response sequence (`CSI
[?] Ps1 ; Ps2 $ y`) into a caller-provided buffer.
## Example
```c
#include <stdio.h>
#include <ghostty/vt.h>
int main() {
char buf[32];
size_t written = 0;
// Query a terminal's cursor visibility and encode a DECRPM report
GhosttyMode mode = GHOSTTY_MODE_CURSOR_VISIBLE;
bool value = false;
ghostty_terminal_mode_get(terminal, mode, &value);
GhosttyModeReportState state = value
? GHOSTTY_MODE_REPORT_SET
: GHOSTTY_MODE_REPORT_RESET;
if (ghostty_mode_report_encode(mode, state, buf, sizeof(buf), &written)
== GHOSTTY_SUCCESS) {
// writes ESC[?25;1$y or ESC[?25;2$y
fwrite(buf, 1, written, stdout);
}
}
```
Add ghostty_mode_report_encode() which encodes a DECRPM response
sequence into a caller-provided buffer. The function takes a mode
tag, a report state integer, an output buffer, and writes the
appropriate CSI sequence (with ? prefix for DEC private modes).
The Zig-side ReportState is a non-exhaustive c_int enum that uses
std.meta.intToEnum for safe conversion to the internal type,
returning GHOSTTY_INVALID_VALUE on overflow. The C header exposes
a GhosttyModeReportState enum with named constants for the five
standard DECRPM state values.
Add modes.h with GhosttyModeTag (uint16_t) matching the Zig ModeTag
packed struct layout, along with inline helpers for constructing and
inspecting mode tags. Provide GHOSTTY_MODE_* macros for all 39
built-in modes (4 ANSI, 35 DEC), parenthesized for safety.
Add ghostty_terminal_mode_get and ghostty_terminal_mode_set to
terminal.h, both returning GhosttyResult so that null terminals
and unknown mode tags return GHOSTTY_INVALID_VALUE. The get function
writes its result through a bool out-parameter.
Add a note in the Zig mode entries reminding developers to update
modes.h when adding new modes.
Add modes.h with GhosttyModeTag, a uint16_t typedef matching the
Zig ModeTag packed struct layout (bits 0-14 for the mode value,
bit 15 for the ANSI flag). Three inline helper functions provide
construction and inspection: ghostty_mode_tag_new,
ghostty_mode_tag_value, and ghostty_mode_tag_ansi.
This extracts our mode reporting from being hardcoded in termio to being
reusable in the existing `terminal.modes` namespace. The goal is to
expose this via the Zig API libghostty (done) and C API (to do later).
This extracts our mode reporting from being hardcoded in termio
to being reusable in the existing `terminal.modes` namespace. The goal
is to expose this via the Zig API libghostty (done) and C API (to do
later).
Add focus event encoding (CSI I / CSI O) to the libghostty-vt public
API, following the same patterns as key and mouse encoding.
The focus Event enum uses lib.Enum for C ABI compatibility. The C API
provides ghostty_focus_encode() which writes into a caller-provided
buffer and returns GHOSTTY_OUT_OF_SPACE with the required size when the
buffer is too small.
Also update key and mouse encoders to return GHOSTTY_OUT_OF_SPACE
instead of GHOSTTY_OUT_OF_MEMORY for buffer-too-small errors, reserving
OUT_OF_MEMORY for actual allocation failures. Update all corresponding
header documentation.
Add focus event encoding (CSI I / CSI O) to the libghostty-vt public
API, following the same patterns as key and mouse encoding.
The focus Event enum uses lib.Enum for C ABI compatibility. The C API
provides ghostty_focus_encode() which writes into a caller-provided
buffer and returns GHOSTTY_OUT_OF_SPACE with the required size when
the buffer is too small.
Also update key and mouse encoders to return GHOSTTY_OUT_OF_SPACE
instead of GHOSTTY_OUT_OF_MEMORY for buffer-too-small errors,
reserving OUT_OF_MEMORY for actual allocation failures. Update all
corresponding header documentation.
