In 7398e9f a regression was introduced which causes Alacritty to crash
on startup since wayland has a keyboard repeat rate thread started
before our logger is initialized.
Since the latest version of time was rather inconvenient to use anyway
and there is no nice solution for this issue other than downgrading the
`time` version again, the time since startup is now logged instead of
the local time.
This should still provide all the relevant information, while getting
rid of an unnecessary dependency. While it would be possible to also
print the delta between log messages, this can be trivially computed so
it has been omitted to skip adding another `Mutex` to the `Logger`
struct.
Since the CI machine is amd64, it is not possible to just execute the
aarch64 binary to test the application. So instead of running `cargo
test`, we just use `cargo build` instead.
Building on the non-release CI instead of just checking also allows us
to make sure linking works properly before a release fails due to it.
Due to unsoundness issues (c.f., time-rs/time#380 and time-rs/time#293),
determining the local timezone can only happen while single-threaded.
Determine the timezone early in startup and apply the offset to the UTC
timestamp before formatting.
When 'ExpandSelection' binding was added only default binding for
RightClick was added, however to expand block selection holding control
when doing a click is required, so this commit adds a binding for
'RMB + Control'.
This resolves an issue where the regex search could loop indefinitely
when the end point was defined in a location containing a fullwidth
character, thus skipping over the end before termination.
Fixes#5753.
This fixes a regression introduced in 8e584099, where block selections
containing the last cell would have the trailing newline stripped and be
joined into one long line on copy.
This commit adds hand rolled drawing of unicode box drawing[1] and
block elements[2] from ranges U+2500 up to U+259f. While using system
font for such characters will look better most of the time, the
characters tend to overlap or not align, so providing builtin font is
the lesser evil here.
[1] - https://www.unicode.org/charts/PDF/U2500.pdf
[2] - https://www.unicode.org/charts/PDF/U2580.pdfFixes#5485.
Before this patch a hint regex ending in the last column would append a
newline, despite this newline being "invisible" to the user. To match
the expected behavior, newlines are trimmed from regex hint matches.
To ensure consistency the simple and semantic selection also do not
include a newline at the end of the copied text anymore.
Fixes#5697.
This adds the ability to pass title and class over IPC via the
create-window subcommand, so users can run only one instance for windows
of different spurposes in the window managers of their choice.
The existing behavior for the new CreateNewWindow actions was to always
pass in their own options, which would discard the existing options
configured on the terminal's PTY config.
To fix this the behavior for CreateNewWindow is now the same as for the
initial window creation, the config values are overwritten conditionally
based on their individual presence in the CLI options.
However all temporary CLI options set on the "master" Alacritty
instance are discarded by all future windows.
Fixes#5659.
The existing PID/FD atomics in alacritty_terminal/src/tty/unix.rs were
shared across all Alacritty windows, causing problem with the new
multiwindow feature.
Instead of sharing these between the different windows, the master FD
and shell PID are now stored on the `window_context`.
Unfortunately this makes spawning new daemons a little more complicated,
having to pass through additional parameters. To ease this a little bit
the helper method `spawn_daemon` has been defined on the
`ActionContext`, making it accessible from most parts of Alacritty's
event loop.
Fixes#5700.
To allow applications spawned by Alacritty to make use of the
shell/foreground process' working directory, it is now set for all new
processes spawned by Alacritty on Unix platforms.
Fixes#5616.
Co-authored-by: Christian Duerr <contact@christianduerr.com>
This patch resolves a behavior that display area doesn't keep track of
content in a scrollback buffer due to display offset reset when viewport
clear is invoked.
This is similar to #5341, but this problem is caused by viewport clear
instead of new outputs to the viewport.
This fixes an issue where the vi cursor would move down one line if it's
positioned at the topmost visible line, while at least partially scrolled
up into history, when new lines are added to the terminal.
This problem is caused by using a display offset of a grid not scrolled
yet when scrolling and determining a new vi cursor position.
Alacritty's `msg create-window` subcommand would previously inherit all
the CLI parameters from the original executable. However not only could
this lead to unexpected behavior, it also prevents multi-window users
from making use of parameters like `-e`, `--working-directory`, or
`--hold`.
This is solved by adding a JSON-based message format to the IPC socket
messages which instructs the Alacritty server on which CLI parameters
should be used to create the new window.
Fixes#5562.
Fixes#5561.
Fixes#5560.
The current completions required a lot of domain-specific knowledge
about each individual shell and their completion functionality. Much of
which is sparsely documented.
While clap does not generate perfect completions, since parameters like
`-e` are missing completions, it does a reasonable job while requiring
no work on writing these completions.
Since access to `cli.rs` isn't possible from the `build.rs`, these
completions aren't always generated on build. Instead a test verifies
that there has been no changes to these completions and provides a
simple code sample for re-generating them. This should provide a simple
solution with minimal overhead.
Previously Alacritty would always initialize only a single terminal
emulator window feeding into the winit event loop, however some
platforms like macOS expect all windows to be spawned by the same
process and this "daemon-mode" can also come with the advantage of
increased memory efficiency.
The event loop has been restructured to handle all window-specific
events only by the event processing context with the associated window
id. This makes it possible to add new terminal windows at any time using
the WindowContext::new function call.
Some preliminary tests have shown that for empty terminals, this reduces
the cost of additional terminal emulators from ~100M to ~6M. However at
this point the robustness of the daemon against issues with individual
terminals has not been refined, making the reliability of this system
questionable.
New windows can be created either by using the new `CreateNewWindow`
action, or with the `alacritty msg create-window` subcommand. The
subcommand sends a message to an IPC socket which Alacritty listens on,
its location can be found in the `ALACRITTY_SOCKET` environment
variable.
Fixes#607.
The existing cursor inversion logic was causing more problems than it
solved, without solving the problem of invisible cursor when inverting a
cell with matching foreground and background colors.
This patch reworks this logic and only inverts the cursor when the
foreground and background colors of the cursor are similar and the
cursor colors aren't set to fixed RGB values.
Fixes#4564.
Fixes#5550.
In this change I went through all current rustfmt configuration options
and expanded our existing configuration with overrides whenever deemed
appropriate.
The `normalize_doc_attributes` option is still unstable, but seems to
work without any issues. Even when passing macros like `include_str!`
that is recognized properly and not normalized. So while this wasn't an
issue anywhere in the code, it should make sure it never will be.
When it comes to imports there are two new major additions. The
`imports_granularity` and `group_imports` options. Both mostly just
incorporate unwritten rules that have existed in Alacritty for a long
time. Unfortunately since `alacritty_terminal` imports in `alacritty`
are supposed to be separate blocks, the `group_imports` option cannot be
used.
While it might be nice to have the full `Passwd` struct around should
anything ever change in the future, there's really no reason why
Alacritty would need this information. So just removing things for now
makes more sense than adding `#[allow(unused)]`.
While the custom merge strategy can make things simpler in some
scenarios, it has repeatedly shown to silently introduce errors that
have to be manually spotted and corrected after the fact.
It's much more reliable and efficient to always manually oversee
potential conflicts, rather than letting them slip through half of the
time and having to contribute follow-up patches.
