Tweak `-Zmacro-stats` measurement.

It currently reports net size, i.e. size(output) - size(input). After
some use I think this is sub-optimal, and it's better to just report
size(output). Because for derive macros the input size is always 1, and
for attribute macros it's almost always 1.

1 files changed, 3 insertions, 3 deletions

diff --git a/src/doc/unstable-book/src/compiler-flags/macro-stats.md b/src/doc/unstable-book/src/compiler-flags/macro-stats.md
index b2622cff057..f3fa69058a7 100644
--- a/src/doc/unstable-book/src/compiler-flags/macro-stats.md
+++ b/src/doc/unstable-book/src/compiler-flags/macro-stats.md
@@ -10,12 +10,12 @@ generated code is normally invisible to the programmer.
 
 This flag helps identify such cases. When enabled, the compiler measures the
 effect on code size of all used macros and prints a table summarizing that
-effect. For each distinct macro, it counts how many times it is used, and the
-net effect on code size (in terms of lines of code, and bytes of code). The
+effect. For each distinct macro, it counts how many times it is used, and how
+much code it produces when expanded (in lines of code, and bytes of code). The
 code size evaluation uses the compiler's internal pretty-printing, and so will
 be independent of the formatting in the original code.
 
-Note that the net effect of a macro may be negative. E.g. the `cfg!` and
+Note that the output size of a macro may be zero. E.g. the `cfg!` and
 `#[test]` macros often strip out code.
 
 If a macro is identified as causing a large increase in code size, it is worth