Related to #1768 but doesn't fix it properly.
This is a temporary hack to avoid some issues with fonts that have mixed
color/non-color glyphs. If there are mixed presentations and the font
does not have emoji codepoints, then we assume it is text. This fixes
the typical scenarios.
This is not a long term solution. A proper long term solution is to
detect this scenario and on a per-glyph basis handle colorization (or
the lack thereof) correctly. It looks like to do this we'll have to
parse some font tables which is considerably more work so I wanted to do
this first.
Fixes#1795
This only affected CoreText. When testing with Freetype the
strikethroughs looked correct for fonts with and without leading
metrics.
This commit adjusts our strikethrough position for fonts that have a
leading metric set to better center it. Previously, we centered the
position _including_ the leading value. The leading value is blank, so
we must center it excluding that value.
Fixes#1784
This was just a misunderstanding of the "spec." When both a y offset
into the image is specified and a height, the image should be stretched.
I mistakingly thought that the image should be offset (even with this
misunderstanding there was a data corruption bug).
This resolves the issue and output matches Kitty.
See: https://github.com/ziglang/zig/issues/17851
Users were noticing that frame render times got slower over time. I
believe (thanks to community for pointing it out) that this is the
culprit.
This works around this issue by clearing and reinitializing the LRU
after a certain number of evictions. When the Zig issue has a better
resolution (either rehash() as a workaround or a better hash
implementation overall) we can change this.
- contains `.elv` file that implements the core of Elvish integration.
- does not contain routines needed for automatic integration.
- stored in `./elvish/lib/...` in preparation for automatic integration:
Elvish imports `.../elvish/lib/*.elv`.
checklist:
- no confirmation on close where the cursor is at prompt:
works, only occasionally doesn't, I'm not yet sure when.
- new terminals start in pwd of previously focused terminal: works
- prompts resize correctly: works
- triple-click while holding `ctrl` selects output of a command:
works (when mouse is over the output)
- cursor at the prompt is turned into a bar: works
- ghostty:`jump_to_prompt` scrolls through prompts: works
- `opt`-click moves cursor at the prompt: works
- `sudo` preserves ghostty terminfo:
untested - not sure when this is needed exactly, but did not encounter
any errors after sudo, either
When the -c option is present, then commands are read from the first
non-option argument command string. Our simple implementation assumes
that if we see at least the '-c' option, a command string was given, and
the shell is always considered to be non-interactive - even if the '-i'
(interactive) option is also given.
Fixes#1409
Fixes macOS 14.4,14.5 kernel panics
As noted in the comment in this diff, I've decided to default this to
true because it fixes a couple _really bad_ issues for macOS users. For
macOS users on 14.4 and 14.5 (latest released as of this commit), this
resolves or at least heavily mitigates a possible kernel panic.
This also fixes#1409 where external displays over certain connections
such as DisplayLink would be unusably slow to render.
Partial fix for link highlight dirty tracking. Some issues remain with
soft-wrapped links (see TODO in code), as well as pressing cmd while
hovering without moving the mouse at all doesn't highlight until you do
move the mouse for some reason.
When hot reloading config with a new font, shaper cache data needs to be
invalidated and the font grid needs to be rebuilt. This change just
makes that happen on all config reloads since it's a rare action so it's
not a performance concern.
Allows for high dpi displays to get odd numbered pixel sizes, for
example, 13.5pt @ 2px/pt for 27px font. This implementation performs
all the sizing calculations with f32, rounding to the nearest pixel
size when it comes to rendering. In the future this can be enhanced
by adding fractional scaling to support fractional pixel sizes.
Now that we're padding the cells with blanks if we have shaped ligatures
we don't actually know the exact count based on the CoreText APIs, so we
should just dynamically add.
