Whenever a hunk refers to an empty line of context, the line usually
consists of a single space and a newline, like this:
@@ -a,b +c,d @@
+foo
<space><newline>
-bar
-baz
However, if for some reason the leading space gets dropped (e.g. when people
send patches through a mailing list software which eats the extra space away,
now in case you are wondering what kind of software would do something
silly like that I will give you a hint: T 1 g r 1 5 dot 0 r g), so that
the above hunk gets transformed to:
@@ -a,b +c,d @@
+foo
<newline>
-bar
-baz
... then 'svn patch' will ignore all lines from the empty line downwards
as noise and only apply the +foo change in our example hunk.
That is, everything leading up to but not including the empty line is applied,
and the rest is ignored.
UNIX patch, however, tolerates this kind of mangling and applies such
patches just fine. So maybe our own unidiff parser should tolerate it, too.
Task
Major