| Age | Commit message (Collapse) | Author |
|---|
|
If a cell has a matching foreground and background and is inversed
through the escape or selection, it will now fall back to the default
background on top of the default foreground.
This makes it possible to show invisible text like this by selecting it.
Hidden text is unaffected by this change.
This fixes #2315.
|
|
|
|
Fixes #34.
Fixes #2012.
|
|
|
|
Since the block cursor inverts the background and foreground colors of a
cell, the hidden cursor has done the same thing without rendering a
cursor since it was using the block cursor shape.
A new `Hidden` cursor style has been introduced for explicitly handling
the invisible cursor differently.
This fixes #2342.
|
|
The ConPTY had a cryptic 'assertion failed: success' error whenever
spawning the shell failed.
This changes the ConPTY assertion to a human-readable error which
clearly states that the file could not be found.
The WinPTY backend has not been fixed yet.
This fixes #2016.
|
|
This fixes a recent regression in
cfc20d4f34dca535654cc32df18e785296af4cc5 which broke cursor colors when
specified in the `colors.cursor` field in the config.
It also removes a lot of unneeded code from the font crate related to
the cursor rendering.
This fixes #2338.
|
|
|
|
Previously cursor dimensions were not calculated correctly when a font
offset was specified, since the font offset was completely ignored.
This has been fixed by moving all the cursor logic from the font into
the Alacritty crate, applying the config's offsets before rasterizing
the cursors.
This has also fixed an issue with some cursors not being rendered as
double-width correctly when over double-width glyphs.
This fixes #2209.
|
|
|
|
This resolves performance issues with the `update_lines` method that
were caused by excessive updates without underlines or strikeout
present.
This also resolves a bug that was causing the underline and strikeout to
extend beyond the end of line in some rare cases.
This fixes #114.
|
|
|
|
Instead of calling TIOCSWINSZ for every pixel change it will now be called only on changes to the grid size. This should reduce screen refreshes.
This fixes #2177.
|
|
If a resize event is identical to the current size, it is no longer
propagated but the resize is discarded immediately.
To further prevent resizes when not necessary, the list of monitors is
enumerated and the DPR of the first display is assumed to be the target
DPR.
This allows spawning a window with dimensions when the config has
columns and lines specified and the window only needs to be resized if
the estimated DPR is not correct.
Fixes #1825.
Fixes #204.
|
|
This fixes the execution of subprocesses on macOS which could sometimes
prevent actions like `SpawnNewProcess` or custom commands from launching
their processes correctly.
This fixes #2259.
|
|
|
|
By setting the minimum Rust version and enforcing it with CI, Alacritty
should hopefully make it possible for maintainers to package the
application even on distributions which are not rolling release.
The 1.31.0 target has been chosen here because it's the first version of
the Rust 2018 release. Bumping this version in the future should be
considered to be a breaking change and should only be done with caution
and in consideration of the supported Rust versions of the major
distributions available.
This fixes #2277.
|
|
|
|
|
|
|
|
This updates winit and glutin to the latest git versions instead of
relying on releases to fix a build failure on BSD.
Unfortunately the upstream version bumps might still take a bit and
there hasn't been any feedback with regards to a planned release date.
This also fixes a slight issue on Windows where selections would start
when focusing a window and a build failure with some Linux systems
Fixes #2243.
Fixes #2170.
Fixes #2244.
|
|
|
|
|
|
This reverts the removal of the call to libc::daemon and instead adds an
annotation to ignore the deprecation warnings on macos.
This will not be an issue in the future since macOS is only discouraging
the use of `daemon`, but I'm not aware of any intention to actually
remove it.
This fixes #2211.
|
|
|
|
This completely reworks URL highlighting to fix two issues which were
caused by the original approach.
The primary issues that were not straight-forward to resolve with the
previous implementation were about handling the URL highlighted content
moving while the highlight is active.
This lead to issues with highlighting with scrolling and when the
display offset was not 0.
The new approach sticks closely to prior art done for the selection,
where the selection is tracked on the grid and updated whenever the
buffer is rotated.
The truncation of URLs was incorrectly assuming input to be just a
single codepoint wide to truncate the end of URLs with unmatching
closing parenthesis. This is now handled properly using Rust's built-in
Unicode support.
This fixes #2231.
This fixes #2225.
|
|
Since the `--class` flag has been changed to take effect on Wayland too,
the documentation has been updated to reflect that.
The original change was made in #2077.
|
|
If the window is resized while lines are longer than the visible area,
Alacritty will no longer move down the prompt and pull from history when
possible but instead keep the prompt in place and move the additional
lines into the scrollback buffer.
This fixes #2213.
|
|
|
|
|
|
The URL highlight stores the state of the last URL highlight with the
position of the URL start position. However when resizing, it's possible
that the indices of this point change which will cause a crash if the
old positions are not within the grid anymore.
This has been resolved by resetting the URL highlight state whenever the
terminal is resized.
The original PR incorrectly required the shift modifier to be required
when the user was in the alternate screen buffer. However the correct
behavior is to require it when the mouse mode is enabled.
This has been resolved and URLs are now highlighted in the alt screen
even if no shift modifier is pressed.
This fixes #2194.
|
|
Since double-width characters are followed by an empty cell containing
only the `WIDE_CELL_SPACER` flag, the URL parser would stop once
encountering the cell after a double-width character.
By skipping cells that contain the `WIDE_CELL_SPACER` flag and
incrementing the URL length by unicode width of the character instead of
cell count, this can be resolved for both URL launching and URL
highlighting.
Fixes #2158.
|
|
Since version 10.5 of macOS the libc::daemon call has been deprecated.
While it is recommended by macOS to use launchd instead, this is not
easily available on other unix platforms.
However since we just spawn a daemon process to prevent Alacritty from
spawning zombies, we can manually invoke `fork` in the child process to
cause a double-fork and re-parent the child process under init so it can
be reaped automatically.
Since the daemon call is not part of POSIX, using the double fork on all
unix platforms also has some portability advantages.
|
|
If a URL ends right at the end of the terminal, it would sometimes
incorrectly include the characters from the following line when
launching the URL.
Similar to the semantic search function, the URL parsing iterator will
now stop if it encounters a cell at the end of the line which does not
contain the `WRAPLINE` flag.
This fixes #1906.
|
|
This changes the cursor whenever it moves to a cell which contains
part of a URL.
When a URL is hovered over, all characters that are recognized as part
of the URL will be underlined and the mouse cursor shape will be
changed. After the cursor leaves the URL, the previous hover state is
restored.
This also changes the behavior when clicking an illegal character right
in front of a URL. Previously this would still launch the URL, but strip
the illegal character. Now these clicks are ignored to make sure there's
no mismatch between underline and legal URL click positions
|
|
This removes all inclusive range code since it has been recently
stabilized in the standard lib.
|
|
This fixes #2010.
|
|
|
|
Fixes #458.
Fixes #1681.
|
|
Alacritty will now automatically reflow lines and shrink them when they
would usually exceed the new width of the terminal instead of
truncation.
If a line had to be truncated, it will also be reflown into the previous
line after growing the terminal width.
The reflow behavior when not at the bottom of the history is similar to
that of VTE and aims to keep the viewport stationary whenever possible.
Opposed to VTE, reflow will also be performed in the alternate screen
buffer.
There will be bugs when resizing the terminal emulator to a size smaller
than the prompt, though these issues were present in all terminal
emulators with reflow support.
This fixes #591.
|
|
|
|
This removes the the signal handling machinery in tty::unix, and
replaces it with functionality from signal-hook, which should be more
robust. Signals caught by signal-hook wake up the existing I/O event
loop, which then delegates back to the PTY to handle them.
In particular, this allows `SIGCHLD` (i.e. child process exits) to shut
down the terminal promptly, instead of sometimes leaving the window
lingering.
Fixes #915.
Fixes #1276.
Fixes #1313.
As a side effect, this fixes a very rare bug on Linux, where a `read`
from the PTY on the master side would sometimes "fail" with `EIO` if the
child closed the client side at a particular moment. This was subject to
a race condition, and was very difficult to trigger in practice.
|
|
|
|
This fixes opening a new instance in the shell's current working
directory. The code currently assumes that there is a Linux
compatible procfs mounted on /proc which is not the case on FreeBSD.
However linprocfs(5) is often mounted on /compat/linux/proc so we
can use that here for the time being.
A proper solution would look up the kern.proc.cwd.<pid> sysctl
instead, but that is not currently easily doable due to missing
KERN_PROC_CWD and struct kinfo_file definitions in the libc crate.
|
|
|
|
In cases where the Alacritty process had invalid std handles then
the ConPTY subprocess would fail to spawn. By setting appropriate
flags we prevent these handles from being passed to the ConPTY
subprocess.
|
|
This fixes #2109.
|
|
This fixes two bugs with the alternate screen buffer.
When resetting while in the alt screen, Alacritty would not swap out
the grids leading to scrollback getting disabled. By swapping out the
grids again when resetting in the alternate screen buffer, scrollback is
now unaffected from a reset.
There was another issue with the cursor jumping around when leaving the
alt screen even though it was not active, this was fixed by skipping all
alt screen swap routines unless the current state matches the expected
state.
This fixes #2145.
|
|
|
|
This fixes two issues with live reloading the `scrolling.history`
property.
It is now possible to increase the scrollback history without restarting
Alacritty.
When decreasing the scrollback history while scrolled beyond the new
history limit, Alacritty will reset the viewport position to the new
limit instead of crashing.
This fixes #2147.
|
