| Age | Commit message (Collapse) | Author |
|---|
|
|
|
Useful when requesting more info to help investigating issues.
|
|
|
|
When RUST_LOG environment variable is set, uses env_logger instead of
our custom logger. This is desirable for debugging purposes.
|
|
This reverts commit e17d38167e174a2cf664e430fe968ec6492e1f08.
Was breaking builds for mac users.
|
|
This PR fixes a few wayland issues of alacritty (and updates glutin on
the process because it is needed).
Mainly two changes are done:
1. Add a drawing_ready() method on Window: see
https://docs.rs/winit/0.8.2/winit/os/unix/trait.WindowExt.html#tymethod.is_ready
for explanations. Hopefully glutin will be able to handle it itself in
the future, but it currently does not.
2. resize window and OpenGL contextes.
The way wayland forces winit to draw its own decorations and how surface
size is defined by its content means that in practice:
- winit's window.set_inner_size() defines the dimensions of the
borders
- glutins gl_window.resize() defines the dimensions of the content
(and is a noop in other platforms)
It is for now glutin's user responsibility to keep them in sync
otherwise borders are drawn stupidly. This PR changes the resize methods
of alacritty::Window to always update both.
This fixed the borders issues for me, tested on weston.
|
|
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.
|
|
|
|
|
|
|
|
|
|
Resolves #679
|
|
* Update to latest Glutin/winit
This *finally* gets us off the fork of Glutin we've been on for so long
and will unblock a number of other items. Functionality should be the
same as before.
The update forced our hand on a compiler update. It's no longer
feasible to pin on an old version. From now on, we require latest
stable.
|
|
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.
|
|
Notable about this implementation is it takes a different approach for
managing cursor cells that previously. The terminal Grid is now
borrowed *immutably*. Instead of mutating Cells in the Grid, a list is
managed within the RenderableCellsIter. The cell at the cursor location
is skipped over, and instead cells are popped off a list of cursor
cells.
It would be good in the future to share some more code between the
different cursor style implementations for populating the cursor cells
list.
Supercedes #349.
|
|
|
|
Support is added for setting _NET_WM_PID automatically. This is to
support scripting of the window environment. For example, this makes it
possible to script opening a window with same CWD:
1. Retrieve the current window
2. (new) get PID of window
3. Check if it's Alacritty, find first child (presumably a shell), and
get the child's cwd.
4. Spawn new instance of terminal with cwd.
Unaddressed in this commit is how this will coexist on a Wayland system.
|
|
OSC strings with UTF-8 previously failed.
|
|
|
|
Resolves #422
|
|
Resolves #211.
|
|
|
|
This passes the vttest for save and restore cursor position. The
implementation was done according to:
http://www.vt100.net/docs/vt510-rm/DECSC.html
As of yet, there are a few things not supported by the terminal which
should otherwise be saved/restored.
vte was updated for a fix with CSI param parsing
|
|
|
|
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.
|
|
|
|
|
|
The `rustc-test` crate has this feature disabled by default causing all
of the test `println!` output to be displayed.
|
|
This uses the rustc-test crate, a copy of the standard test crate, to
dynamically create tests for each reference test. No need to remember to
update the macro, just add the directory to ref!
|
|
Resolves #23
Resolves #144
|
|
|
|
|
|
|
|
Resolves #35.
|
|
All of the changes in this commit are due to clippy lints.
|
|
|
|
Also adds a feature `err-println` for enabling `err_println!` printing.
|
|
Ref tests use a recording of the terminal protocol and a serialization
of the grid state to check that the parsing and action handling systems
produce the correct result. Ref tests may be recorded by running
alacritty with `--ref-test` and closing the terminal by using the window
"X" button. At that point, the recording is fully written to disk, and a
serialization of important state is recorded. Those files should be
moved to an appropriate folder in the `tests/ref/` tree, and the
`ref_test!` macro invocation should be updated accordingly.
A couple of changes were necessary to make this work:
* Ref tests shouldn't create a pty; the pty was refactored out of the
`Term` type.
* Repeatable lines/cols were needed; on startup, the terminal is resized
* by default to 80x24 though that may be changed by passing
`--dimensions w h`.
* Calculating window size based on desired rows/columns and font metrics
required making load_font callable multiple times.
* Refactor types into library crate so they may be imported in an
integration test.
* A whole bunch of types needed symmetric serialization and
deserialization. Mostly this was just adding derives, but the custom
deserialization of Rgb had to change to a deserialize_with function.
This initially adds one ref test as a sanity check, and more will be
added in subsequent commits. This initial ref tests just starts the
terminal and runs `ll`.
|
|
Committed this on a plane with no internet; need to get a real glutin
ref pushed somewhere and update this commit before merging into master.
|
|
Which means it can be disabled in release builds. No more working on a
renderer feature and actually breaking the Alacritty your editor is
running inside.
|
|
Now uses serde_dervive \o/
|
|
Resolves #10.
|
|
Currently it only supports x11 via the xclip program, and that only
supports reading the clipboard contents.
|
|
doesn't work on ubuntu 16.04 for some reason
|
|
This was largely an experiment to see whether writing and reading from a
separate thread was causing terminal state corruption as described in
https://github.com/jwilm/alacritty/issues/9. Although this doesn't seem
to fix that particular issue.
Keeping this because it generally seems more correct than
reading/writing from separate locations.
|
|
Also enables debug symbols in release profile by default. Until
Alacritty ships, there's going to be lots of perf analysis which needs
debug symbols.
The PriorityMutex low priority method was never used. Now it's just a
fair mutex.
|
|
|
|
Using the vte crate allows removal of the ansi parser state machine and
enables us to just be concerned with actions described in the protocol.
In addition to making alacritty simpler, this also improves correctness
and performance.
|
|
Currently has a bug where screen is blank at startup. That aside,
Alacritty uses basically 0 CPU now. The input thread is still separate
from the render thread, but, given the ability to wake the event loop,
it may be possible to merge them again. I'm not sure if that's actually
desirable.
Performance is seemingly unchanged.
|
