| Age | Commit message (Collapse) | Author |
|---|
|
The extra window padding was calculated in the renderer which lead to
problems with the paddings calculated in the `src/display.rs` and
`src/term/mod.rs`.
As a solution, every instance of `config.padding().x/y` has been removed
from the renderer (`src/renderer/mod.rs`), instead the padding is always
passed through from the `src/display.rs`.
The initial calculations during display creation and after resize then
are scaled appropriately and then the extra padding is calculated. As a
result every other location can just make use of the correctly
calculated `size_info.padding_x` and `size_info.padding_y`.
The documentation has been changed to clearly state that the padding is
scaled by DPI now.
This fixes #1773.
|
|
Currently alacritty always puts the grid at the top-left position of the
window. The only distance to the top-left window border is set by the
padding in the config.
However the grid always has a fixed size, and if a cell doesn't
completely fit the screen anymore, the padding at the bottom right
window corner can be significantly bigger than the padding at the top
left.
To fix this whenever there is more space left and there would usually be
a bigger padding at the bottom right, the space is now split up and
added to the padding.
This should always center the grid inside the window and make sure all
borders have the same padding from the text area.
This screenshot shows how it has been until now:
Here is how it looks now:
This fixes #1065.
|
|
|
|
Some changes include:
• Use the with_hardware_acceleration function on the ContextBuilder to not require the discrete GPU
• Remove the LMenu and RMenu virtual key codes (winit 0.16.0 removed these because Windows now generates LAlt and RAlt instead
• Replace set_cursor_state with hide_cursor (winit 0.16.0 removed the set_cursor_state function)
• Replace GlWindow::hidpi_factor with GlWindow::get_hidpi_factor and change to expecting an f64
• Use the glutin/winit dpi size and position types where possible
Glutin's dpi change event has been implemented. All size events now
return logical sizes. As a result of that, the logical sizes are translated
in the `display::handle_rezize` method so DPI scaling works correctly.
When the DPI is changed, the glyph cache is updated to make use of the
correct font size again.
Moving a window to a different screen which is a different DPI caused a
racing condition where the logical size of the event was sent to the
`handle_resize` method in `src/display.rs`, however if there was a DPI
change event before `handle_resize` is able to process this message, it
would incorrectly use the new DPI to scale the resize event.
To solve this issue instead of sending the logical size to the
`handle_resize` method and then converting it to a physical size in
there, the `LogicalSize` of the resize event is transformed into a
`PhysicalSize` as soon as it's received. This fixes potential racing
conditions since all events are processed in order.
The padding has been changed so it's also scaled by DPR.
The `scale_with_dpi` config option has been removed. If it's not present
a warning will be emitted.
The `winit` dependency on Windows has been removed. All interactions
with winit in Alacritty are handled through glutin.
|
|
Initial support for Windows is implemented using the winpty translation
layer. Clipboard support for Windows is provided through the `clipboard`
crate, and font rasterization is provided by RustType.
The tty.rs file has been split into OS-specific files to separate
standard pty handling from the winpty implementation.
Several binary components are fetched via build script on windows
including libclang and winpty. These could be integrated more directly
in the future either by building those dependencies as part of the
Alacritty build process or by leveraging git lfs to store the artifacts.
Fixes #28.
|
|
The terminal lock is now dropped before rendering by storing
all grid cells before clearing the screen.
This frees the terminal to do other things since the lock is now
free, which lead to a performance benefit with high throughput
applications.
|
|
This adds support for the `hidden` escape sequence `\e[8m`, which will
render the text as invisible.
This has also raised a few questions about the rendering of foreground
and background colors and their interaction with the different escape
sequences. Previously, Alacritty has oriented itself after URxvt, which
has some strange and unexpected behavior.
The new implementation of color inversion is modeled after XTerm, which
has a consistent pattern of always inverting the foreground and
background colors. This should hopefully lead to less confusion for the
user and a more consistent behavior.
A full matrix showcasing the new way Alacritty inverses text can be
found here:
https://i.imgur.com/d1XavG7.png
This fixes #1454 and fixes #1455.
|
|
* Change deb installation from crates.io to git
There have been a number of issues an PRs opened since
the cargo-deb installation does not work with the latest
version from crates.io.
To help out users until the crates.io version is updated,
the installation instructions have been temporarily
changed to install `cargo-deb` through github.
* Revert cargo-deb install back to use crates.io
Since `cargo-deb` has been updated on crates.io it is now
possible to just install it from crates.io and build Alacritty's
deb without having to rely on github.
* Update dependencies
This fixes an `illegal hardware instruction (core dumped)`
error when building in release mode.
* Remove redundant copy when selecting font_key
* Bump version number to 0.2.0
Since the Scrollback branch introduces some major changes, this bumps
the version number from 0.1.0 to 0.2.0.
The versions of Alacritty have not been updated regularly to this point,
so the scrollback branch is a good point in time to start updating
Alacritty's version on a regular basis.
Further changes to the readme, like dropping the 'alpha' status and
updating it to 'beta' could also be introduced with this branch. This
way there will be a clean cut which updates everything as soon as
scrollback is merged.
Building versions is another thing which would be a good thing to start
reasonably quickly. However starting this on the main branch after
scrollback has been merged seems like a more reliable way to move
forward.
This fixes #1240.
* Add a CHANGELOG file
A CHANGELOG file has been added to offer a bit more transparency over
which features have been changed, added and potentially removed in
Alacritty.
There are various formats available for the CHANGELOG file but the most
common and sensible one seems to be the one defined by
https://keepachangelog.com/en/1.0.0. Following the template proposed by
this it should be possible to create a clear CHANGELOG which makes it
simple for new contributors to figure out exactly which formatting
should be used for it.
Since there have been quite a few changes to Alacritty already, not all
changes have been added to the changelog. However a few entries have
been ported just to give a bit of an example what the format should look
like. This also helps with the 0.2.0 version since it will not be
completely empty in the changelog.
This fixes #1534.
* Update CHANGELOG
This updates the CHANGELOG to include the changes introduced by
43882ade33d4c14ee7248e489a2d33395faaa0b1.
|
|
There were some unneeded codeblocks and TODO/XXX comments in the code
that have been removed. All issues marked with TODO/XXX have either been
already resolved or tracking issues exist.
|
|
|
|
|
|
Prevent the cell dimensions from going below 1, this bug resulted in
allocation of large amounts of memory in the scrollback PR but is also
present on master.
Currently the approach is to just `panic!`, however an `eprintln!` and
`exit` could be an alternative too. I don't think it's realistic to
check this at startup and it should have no performance impact since the
failing method is only called once at startup.
To make it a bit more clear what kind of values are accepted, the
datatypes of offsets and paddings have also been changed so that these
don't accept floats anymore and padding can never be negative.
This should allow us to be a bit more strict with the config to make
sure that errors are printed when invalid values are specified (like
negative padding).
This fixes #1167.
|
|
Unwrapping inside the config file parsing can lead to some issues that
prevent us from falling back to a default configuration file.
One instance of that issue was mentioned in #1135.
Now all instances of `unwrap()` have been removed and replaced with
proper error handling. This will make the config more robust and
prevents live reload from silently breaking while alacritty is running.
This also fixes a few currently existing clippy issues.
Clippy added an additonal lint which complains about `MyStruct { field:
field }`.
These issues have been fixed, except for some false-positives and issues
in external macros which will probably be fixed with future updates (rust-lang-nursery/bitflags#149)
|
|
Updated the version of some dependencies.
This also changes to a new clippy version so clippy can work with the latest nightly compiler again. Some issues created by new lints have been fixed.
|
|
This commit adds clippy as a required step of the build process. To make
this possible, all existing clippy issues have been resolved.
|
|
This replaces the `font_size_modifier` stored on the `Term` struct with a `font_size` field.
With this change it is not necessary anymore to calculate the new font size from a delta but the current font size is always stored directly on the `Term` struct.
As a result of this it is now possible to increase the font size by more than 127 steps at runtime. It also limits the minimum font size to 1, so issues with the `font_size_modifier` dropping far below font size 1 are resolved with this change.
This fixes #955.
|
|
This resolves the remaining issue of #842.
|
|
|
|
Previously there were two separate but intended-to-be-identical
implementations. Now the two implementations simply delegate to a
single, shared method. This should help correctness issues in the
future.
|
|
|
|
When an atlas is full and the `insert` call fails, a new atlas should be
created. This is the current strategy, however the atlas is put at the
end of the vector, but the `current_atlas` index is set to 0, instead of
the last element. This leads to a recursion where it keeps trying to
insert into the full atlas at position 0 instead of the new at
`atlas.len() - 1`.
With this simple fix a stack-overflow is prevented because the new atlas
is inserted as the first element, which then will be used correctly for
loading new glyphs.
This fixes jwilm/alacritty/issues/842 and might also solve
jwilm/alacritty/issues/914 (I wasn't able to reproduce this with the
latest master).
|
|
(needless_pass_by_value, needless_borrow).
|
|
to "==".
|
|
Resolves #872
|
|
Since landing the patch adding transparency support to Alacritty,
there's been an issue where othewise solid background cells were also
being rendered partially transparent. Now, all filled background cells
are rendered fully opaque.
Some logic was added to support discarding filled backgrounds which had
the same color as the default background. This means that, if the
default background is #000 and a cell has that background, it will never
be rendered opaque. This may not be correct.
Note that many truecolor vim color schemes print spaces for default
colored background cells. Performance can be dramatically improved by
using ctermbg=NONE guibg=NONE to skip rendering those cells.
|
|
The source of the leak was loading up multiple copies of the FT_face
even when not necessary. Fonts are now appropriately cached for
FreeType when going through the `Rasterize::load_font` API.
Additionally, textures in the glyph cache are now reused.
The result of this is that resizing to already loaded fonts is free
from a memory consumption perspective.
|
|
Adds support for font resizing at run-time. Three new actions are
introduced:
* IncreaseFontSize - Increases current font size by 1.0
* DecreaseFontSize - Decreases current font size by 1.0
* ResetFontSize - Resets font size to that specified in the
configuration.
The stock config files have example configuration for each which should
match gnome-terminal. For convenience, the config entries are:
- { key: Key0, mods: Control, action: ResetFontSize }
- { key: Equals, mods: Control, action: IncreaseFontSize }
- { key: Subtract, mods: Control, action: DecreaseFontSize }
|
|
Because there are so many clippy warnings in the current codebase,
this commit removes '#![cfg_attr(feature = "clippy", deny(clippy))]',
to make it easier to fix warnings incrementally.
|
|
|
|
500ms introduced a visual lag between file save and display update.
|
|
|
|
The option is an Alpha struct that ensures that the contained float is
between 0.0 and 1.0. Background colors are multiplied by the opacity
to properly alpha blend them.
|
|
|
|
This reverts commit 3cdba291242cc1c1684bff7f8242262b1cdeb582.
On some systems, the target commit actually caused a massive performance
issue rather than fixing one.
|
|
The InstanceData type in the rendering subsystem was previously 16
floats which occupied a total of 64 bytes per instance. This meant that
for every character or background cell drawn, 64 bytes were sent to the
GPU. In the case of a 400x100 cell grid, a total of 2.5MB would be sent.
This patch reduces InstanceData's size to 26 bytes, a 60% improvement!
Using the above example for comparison, a worst case of 1MB would be
transferred.
The motivation for this patch comes from macOS. Once the terminal grid
would reach a certain size, performance experienced a sharp and dramatic
drop (render times would go from ~3ms to ~16ms). I don't want to
speculate too much on the underlying issue, but suffice it to say that
this patch alleviates the problem in my testing.
|
|
|
|
|
|
|
|
Padding can be configured by using the `padding` field in the config
file, like so:
padding:
x: 2
y: 2
which would result in a 2px padding within each side of the window.
|
|
We previously had a hard-coded value for aligning glyphs within cells.
The font descent is now used, and the offset should be correct by
default.
|
|
The two structs are very similar, so there is no reason for them to be
separate. Instead combine them into a single Delta struct, which can be
used to shift a point in a two dimensional plane.
|
|
Add the ability to move glyphs within their cells on a global basis via
an option in the configuration file.
|
|
The changes to metric consumption rendered the size argument
unnecessary, remove it.
|
|
The color list needs to be updated by the parser, and this isn't
possible if it's on the config. This change makes sense semantically as
well since it's really part of the terminal state.
This is in preparation for OSC color parsing.
|
|
This commit adds support for a visual bell. Although the Handler in src/ansi.rs
warns "Hopefully this is never implemented", I wanted to give it a try. A new
config option is added, `visual_bell`, which sets the `duration` and `animation`
function of the visual bell. The default `duration` is 150 ms, and the default
`animation` is `EaseOutExpo`. To disable the visual bell, set its duration to 0.
The visual bell is modeled by VisualBell in src/term/mod.rs. It has a method to
ring the bell, `ring`, and another method, `intensity`. Both return the
"intensity" of the bell, which ramps down from 1.0 to 0.0 at a rate set by
`duration` and `animation`.
Whether or not the Processor waits for events is now configurable in order to
allow for smooth drawing of the visual bell.
|
|
Loading a glyph from the cache is a very hot operation in the renderer.
The original implementation would first check if a glyph was loaded and
then call `get()` which would have to search a second time. This showed
up as a very slow point in profiles.
This patch addresses glyph cache access in two ways: by using a faster
hasher optimized for small keys (fnv), and by using the entry API for
fetching a cached glyph. The `fnv` hasher is faster than the default and
is very efficient for small keys. Using the entry API on the HashMap
means only 1 lookup instead of two. The entry API has a downside where
the key needs to get cloned on fetches.
Reducing the GlyphKey width to 64-bits helps in both areas. Copying an
8-byte wide type is very cheap and thus limits downside of the entry
API. The small width also helps with the hasher performance.
Over all, this patch reduced typical render times by several hundred
microseconds on a 2013 MacBook Pro with a full screen terminal full of
text.
|
|
|
|
|
|
|
|
This work started because we wanted to be able to simply say "monospace"
on Linux and have it give us some sort of font. The config format for
fonts changed to accomodate this new paradigm. As a result, italic and
bold can have different families from the normal (roman) face.
The fontconfig based font resolution probably works a lot better than
the CoreText version at this point. With CoreText, we simply iterate
over fonts and check it they match the requested properties. What's
worse is that the CoreText version requires a valid family. With
fontconfig, it will just provide the closest matching thing and use it
(unless a specific style is requested).
|
