Since the expansion of the selection was done after clamping it to the
grid, the selection would incorrectly move the clamped start over by one
cell when the start was to the right of the original column. By
resetting the side of the start point to `Left` before expanding, this
can be circumvented.
This also resolves a regression which broke backwards bracket selection.
Fixes#3223.
Since the assumption is usually that bold text is drawn in bright
colors, this might break some applications. However some other terminals
have already taken this leap, which should lessen the impact for
Alacritty.
Since this might still be desired and necessary for certain
applications, the config option is just switched to draw with normal
colors by default, however the old behavior can still be restored.
Fixes#2779.
This change includes dynamic escape buffer support in VTE, which allows
us to have arbitrary escape sizes.
Since tmux could potentially use very long escapes for the clipboard
escape, this allows copying more text.
Fixes#1002.
This fixes various outdated links pointing to the old jwilm/alacritty
repository.
Since `copypasta` now has its own github repository at
https://github.com/alacritty/copypasta, the sources have been removed
from Alacritty.
This resolves the issue with full width glyphs getting rendered in the
last column. Since they need at least two glyphs, it is not possible to
properly render them in the last column.
Instead of rendering half of the glyph in the last column, with the
other half cut off, an additional spacer is now inserted before the wide
glyph. This means that the specific glyph in question is then three
cells wide.
Fixes#2385.
This fixes a regression introduced in
e99057b179, which used `self.len` to
calculate the remainder of `self.zero` during rotation instead of
`self.inner.len()`, leading to a broken `self.zero` offset and incorrect
rotation.
This fixes a bug where a row would always get reset completely if its
background does not equal the default terminal background. This leads to
big performance bottlenecks when running commands like `echo "\e[41m" &&
yes`.
Instead of resetting the entire row whenever the template cell is not
empty, the template cell is now compared to the last cell in the row.
The last cell will always be equal to the previous template cell when
`row.occ < row.inner.len()` and if `occ` is equal to the row's length,
the entire row is always reset anyways.
Fixes#2989.