actually, remove normalization entirely

normalization has some weird edge cases that cause incorrect behavior,
since we aren't implementing full grapheme segmentation (for instance, a
single codepoint can be normalized into three different codepoints, and
there are codepoints that are combining characters but have width != 0)

1 files changed, 0 insertions, 4 deletions

diff --git a/Cargo.toml b/Cargo.toml
index 9e60834..14d7774 100644
--- a/Cargo.toml
+++ b/Cargo.toml
@@ -12,14 +12,10 @@ keywords = ["terminal", "vt100"]
 categories = ["command-line-interface", "encoding"]
 license = "MIT"
 
-[features]
-default = ["unicode-normalization"]
-
 [dependencies]
 itoa = "0.4"
 enumset = "0.4"
 log = "0.4"
-unicode-normalization = { version = "0.1", optional = true }
 unicode-width = "0.1"
 vte = "0.3"