File 0101-kernel-Fix-broken-wcwidth-for-wide-chars.p... of Package erlang

Overview Repositories Revisions Requests Users Attributes Meta

File 0101-kernel-Fix-broken-wcwidth-for-wide-chars.patch of Package erlang

From 0e32395bc63a92ec6d64690ef11060d165317214 Mon Sep 17 00:00:00 2001
From: Lukas Larsson <lukas@erlang.org>
Date: Mon, 15 Apr 2024 20:47:20 +0200
Subject: [PATCH] kernel: Fix broken wcwidth for wide chars

On OSs with old libc, the wcwidth will return -1 for most/all emoji characters
even if they are known to be wide characters. So we add a check that if -1
is returned for something that is known to be wide, we return 2 in size instead
of 0.
---
 lib/kernel/src/prim_tty.erl | 9 ++++++++-
 1 file changed, 8 insertions(+), 1 deletion(-)

diff --git a/lib/kernel/src/prim_tty.erl b/lib/kernel/src/prim_tty.erl
index 3e61207b1e..0fe13b84b8 100644
--- a/lib/kernel/src/prim_tty.erl
+++ b/lib/kernel/src/prim_tty.erl
@@ -1113,7 +1113,14 @@ npwcwidth(Char) ->
     npwcwidth(Char, true).
 npwcwidth(Char, true) ->
     case wcwidth(Char) of
-        {error, not_printable} -> 0;
+        {error, not_printable} ->
+            %% Sometimes libc wcwidth if wrong and returns -1 aka not_printable
+            %% for something that is a wide character. So we try to make things
+            %% better by returning 2 for such characters.
+            case unicode_util:is_wide(Char) of
+                true -> 2;
+                false -> 0
+            end;
         {error, enotsup} ->
             case unicode_util:is_wide(Char) of
                 true -> 2;
-- 
2.35.3

Places

File 0101-kernel-Fix-broken-wcwidth-for-wide-chars.patch of Package erlang

Places