Consider escaping special characters like 'less' does

[Joris Valette <joris.valette@gmail.com>]

The pager 'less' escapes some characters when calling 'git diff'. This
is what I might get:

$ git diff --cached
diff --git a/some_file b/some_file
new file mode 100644
index 0000000..357323f
--- /dev/null
+++ b/some_file
@@ -0,0 +1 @@
+<U+FEFF>Hello
\ No newline at end of file

This example is a simple file encoded in UTF-8 *with BOM*.
On the other hand, the built-in git output shows this:

$ git --no-pager diff --cached
diff --git a/some_file b/some_file
new file mode 100644
index 0000000..357323f
--- /dev/null
+++ b/some_file
@@ -0,0 +1 @@
+Hello
\ No newline at end of file




You can see 'less' shows <U+FEFF> explicitly, while it is implicit and
hidden with git's built-in output.
Other characters behave the same, like ZERO WIDTH SPACE <U+200B>, and
I believe many others.
This is particularly annoying when there are other changes on the same
line. Here I add ' world!' but also remove the BOM:

$ git diff
diff --git a/some_file b/some_file
index 357323f..6769dd6 100644
--- a/some_file
+++ b/some_file
@@ -1 +1 @@
-<U+FEFF>Hello
\ No newline at end of file
+Hello world!
\ No newline at end of file

Compare with:

$ git --no-pager diff
diff --git a/some_file b/some_file
index 357323f..6769dd6 100644
--- a/some_file
+++ b/some_file
@@ -1 +1 @@
-Hello
\ No newline at end of file
+Hello world!
\ No newline at end of file




In the first example I can see both changes, in the second example the
BOM removal is hidden and I won't notice it because there is indeed
another genuine change on the same line.

I'll add a third example with CRLF line-terminators:

$ git diff
diff --git a/some_file b/some_file
index 357323f..dfd6895 100644
--- a/some_file
+++ b/some_file
@@ -1 +1 @@
-<U+FEFF>Hello
\ No newline at end of file
+Hello world!^M

and:

$ git --no-pager diff
diff --git a/some_file b/some_file
index 357323f..dfd6895 100644
--- a/some_file
+++ b/some_file
@@ -1 +1 @@
-Hello
\ No newline at end of file
+Hello world!




My thoughts are that git should add this feature and tend towards
less's behavior. A diff output should show as much as possible.
I would happily use 'less' all the time, but unfortunately I don't
think I can. For instance, 'git add --patch', which I use quite often,
prints through git's built-in output, and I miss less's features.

Joris Valette.