Closes#18263
Improvements:
• **Batch text rendering** - Combine adjacent cells with identical
styling into single text runs to reduce draw calls
• **Throttle hyperlink searches** - Limit hyperlink detection to every
100ms or when mouse moves >5px to reduce CPU usage
• **Pre-allocate collections** - Use `Vec::with_capacity()` for cells,
runs, and regions to minimize reallocations
• **Optimize background regions** - Merge adjacent background rectangles
to reduce number of draw operations
• **Cache selection text** - Only compute terminal selection string when
selection exists
Release Notes:
- Improved terminal rendering performance.
---------
Co-authored-by: Conrad Irwin <conrad.irwin@gmail.com>
This glyphs field is usually larger than 8 elements, and SmallVec is not
efficient when it cannot store the value inline.
This change also adds precise glyphs run preallocation in some places
`ShapedRun` is constructed.
Release Notes:
- N/A
This commit is all about strings, not about line layout at all. When
laying out text, we use a line layout cache to avoid roundtrips to
system layout engine where possible. This makes it so that we might end
up not needing an owned version of text to insert into the cache, as we
might get a cached version.
The API boundary of line layout accepted text to be laid out as &str. It
then performed cache lookup (which didn't require having an owned
version) and only resorted to making an owned version when needed. As it
turned out though, exact cache hits are quite rare and we end up needing
owned version more often than not. The callers of line layout either
dealt with SharedStrings or owned Strings. Due to coercing them into
&str, we were ~always copying text into a new string (unless there was a
same-frame-hit). This is a bit wasteful, thus this PR generifies the API
a bit to make it easier to reuse existing string allocations if there
are any.
Benchmark scenario: scrolling down page-by-page through editor_tests (I
ran the same scenario twice):
Release Notes:
- N/A
Closes https://github.com/zed-industries/zed/issues/4351
Applies the same splitting mechanism, as Zed's central pane has, to the
terminal panel.
Similar navigation, splitting and (de)serialization capabilities are
supported.
Notable caveats:
* zooming keeps the terminal splits' ratio, rather expanding the
terminal pane
* on macOs, central panel is split with `cmd-k up/down/etc.` but `cmd-k`
is a "standard" terminal clearing keybinding on macOS, so terminal panel
splitting is done via `ctrl-k up/down/etc.`
* task terminals are "split" into regular terminals, and also not
persisted (same as currently in the terminal)
Seems ok for the initial version, we can revisit and polish things
later.
Release Notes:
- Added the ability to split the terminal panel
Turns out that in the case you have a somehow valid utf-8 file that
contains almost all ascii control characters, we run out of element
arena space.
Fixes: #20652
Release Notes:
- Fixed a crash when opening a file containing a very large number of
ascii control characters on one line.
This pull request replaces the static `⋯` character we used to insert
when folding a range with a custom render function that return an
`AnyElement`. We plan to use this in the assistant, but for now this
should be behavior-preserving.
Release Notes:
- N/A
---------
Co-authored-by: Nathan <nathan@zed.dev>
Co-authored-by: Conrad <conrad@zed.dev>
This pull request introduces a new `markdown` crate which is capable of
parsing and rendering a Markdown source. One of the key additions is
that it enables text selection within a `Markdown` view. Eventually,
this will replace `RichText` but for now the goal is to use it in the
assistant revamped assistant in the spirit of making progress.
<img width="711" alt="image"
src="https://github.com/zed-industries/zed/assets/482957/b56c777b-e57c-42f9-95c1-3ada22f63a69">
Note that this pull request doesn't yet use the new markdown renderer in
`assistant2`. This is because we need to modify the assistant before
slotting in the new renderer and I wanted to merge this independently of
those changes.
Release Notes:
- N/A
---------
Co-authored-by: Nathan Sobo <nathan@zed.dev>
Co-authored-by: Conrad <conrad@zed.dev>
Co-authored-by: Alp <akeles@umd.edu>
Co-authored-by: Zachiah Sawyer <zachiah@proton.me>
This pull request introduces the new
`ElementContext::request_autoscroll(bounds)` and
`ElementContext::take_autoscroll()` methods in GPUI. These new APIs
enable container elements such as `List` to change their scroll position
if one of their children requested an autoscroll. We plan to use this in
the revamped assistant.
As a drive-by, we also:
- Renamed `Element::before_layout` to `Element::request_layout`
- Renamed `Element::after_layout` to `Element::prepaint`
- Introduced a new `List::splice_focusable` method to splice focusable
elements into the list, which enables rendering offscreen elements that
are focused.
Release Notes:
- N/A
---------
Co-authored-by: Nathan <nathan@zed.dev>
See https://zed.dev/channel/gpui-536
Fixes https://github.com/zed-industries/zed/issues/9010
Fixes https://github.com/zed-industries/zed/issues/8883
Fixes https://github.com/zed-industries/zed/issues/8640
Fixes https://github.com/zed-industries/zed/issues/8598
Fixes https://github.com/zed-industries/zed/issues/8579
Fixes https://github.com/zed-industries/zed/issues/8363
Fixes https://github.com/zed-industries/zed/issues/8207
### Problem
After transitioning Zed to GPUI 2, we started noticing that interacting
with the mouse on many UI elements would lead to a pretty annoying
flicker. The main issue with the old approach was that hover state was
calculated based on the previous frame. That is, when computing whether
a given element was hovered in the current frame, we would use
information about the same element in the previous frame.
However, inspecting the previous frame tells us very little about what
should be hovered in the current frame, as elements in the current frame
may have changed significantly.
### Solution
This pull request's main contribution is the introduction of a new
`after_layout` phase when redrawing the window. The key idea is that
we'll give every element a chance to register a hitbox (see
`ElementContext::insert_hitbox`) before painting anything. Then, during
the `paint` phase, elements can determine whether they're the topmost
and draw their hover state accordingly.
We are also removing the ability to give an arbitrary z-index to
elements. Instead, we will follow the much simpler painter's algorithm.
That is, an element that gets painted after will be drawn on top of an
element that got painted earlier. Elements can still escape their
current "stacking context" by using the new `ElementContext::defer_draw`
method (see `Overlay` for an example). Elements drawn using this method
will still be logically considered as being children of their original
parent (for keybinding, focus and cache invalidation purposes) but their
layout and paint passes will be deferred until the currently-drawn
element is done.
With these changes we also reworked geometry batching within the
`Scene`. The new approach uses an AABB tree to determine geometry
occlusion, which allows the GPU to render non-overlapping geometry in
parallel.
### Performance
Performance is slightly better than on `main` even though this new
approach is more correct and we're maintaining an extra data structure
(the AABB tree).
Release Notes:
- Fixed a bug that was causing popovers to flicker.
---------
Co-authored-by: Nathan Sobo <nathan@zed.dev>
Co-authored-by: Thorsten <thorsten@zed.dev>
Reverts zed-industries/zed#7481
This would regress performance because we'd be using the standard
library's hash maps everywhere, so reverting for now.
