mirrors/foot
1
0
Fork 0
mirror of https://codeberg.org/dnkl/foot.git synced 2026-04-28 06:46:38 -04:00
Code Issues Projects Releases Packages Wiki Activity Actions

config: add grapheme-width-method=default

Browse source 
This is the new default mode, and tries to do "the right thing". That
is, recognize and handle:

* Variation Selector 16 (graphical presentation)
* ZWJ sequences
* Skin tones and hair-styles
* Country flags

Things known to be incorrect:

* "Invalid" country letter combinations are still treated as
  flags. That is, the combined width is 2, even though it should be 4.
This commit is contained in:
Daniel Eklöf 2022-12-14 14:35:00 +01:00
parent 3b9aca6a3d
commit ab39ed99df
No known key found for this signature in database
GPG key ID: 5BBD4992C116573F
4 changed files with 80 additions and 11 deletions
Download patch file Download diff file Expand all files Collapse all files

60
doc/foot.ini.5.scd
View file

@ -1229,21 +1229,63 @@ any of these options.
*grapheme-width-method*
Selects which method to use when calculating the width
(i.e. number of columns) of a grapheme cluster. One of
*wcswidth*, *double-width* and *max*.
(i.e. number of columns) of a grapheme cluster. One of *default*,
*wcswidth*, *double-width* and *max*. See details below.
*wcswidth* simply adds together the individual width of all
codepoints making up the cluster.
Background:
Glyphs rendered in the terminal may be 1 or 2 columns wide. The
terminal emulator and the application running in the terminal must
agree what that width is, or you will see cursor
de-synchronization. This typically manifests itself as rendering
glitches and the cursor "jumping" when moving it over problematic
characters.
Most glyphs consists of a single Unicode codepoint. In this case,
there is usually no issues.
Some glyphs however consists of multiple codepoints. These are
called grapheme clusters, and can often be problematic.
While there is a standard function to calculate the width of a
single codepoint (*wcwidth*(3p)), there is no such function for
calculating the width of a grapheme cluster.
The closest thing we have is *wcswidth*(3p), that calculates the
width of a string. It is usually implemented by adding together
the widths of all codepoints that make up the string. This is not
correct for grapheme clusters. Take 👩🏾‍🚀for example (WOMAN
ASTRONAUT: MEDIUM-DARK SKIN TONE). It consists of 4 codepoints:
WOMAN + MEDIUM-DARK SKIN TONE + ZWJ + ROCKET. The sum of the
widths is 6.
Some applications are better at calculating correct grapheme
cluster widths than other. The *grapheme-width-method* option lets
you choose the method that works best for your use case.
*default* attempts to do "the right thing". When used with
applications that does not handle Unicode grapheme clusters
correctly, you will see cursor de-synchronization.
*wcswidth* simply adds together the individual widths of all
codepoints making up the cluster. This is typically the most
compatible mode, at least for "older" applications.
It will allocate too few cells for legacy emojis with an explicit
variation selector.
It will allocate too many cells for complex emojis consisting of
multiple codepoints.
*double-width* does the same, but limits the maximum number of
columns to 2. This is more correct, but may break some
applications since applications typically use *wcswidth*(3)
internally to calculate the width. This results in cursor
de-synchronization issues.
columns to 2. This is usually more correct, but like *default*,
will result in cursor de-synchronization when used with
applications that are not Unicode grapheme cluster
aware.
*max* uses the width of the largest codepoint in the cluster.
Default: _wcswidth_
Default: _default_
*font-monospace-warn*
Boolean. When enabled, foot will use heuristics to try to verify