Linus Torvalds <torvalds@xxxxxxxxxxxxxxxxxxxx> writes:
> I bisected it, and this bug was introduced almost two years ago. In commit
> 3b6121f69b2 ("git-mailinfo: use strbuf's instead of fixed buffers"), to be
> exact. I'm pretty sure the bug is that handle_commit_msg() was changed to
> use 'strbuf_ltrim()' for the 'still_looking' case.
>
> Before commit 3b6121f69b2, it would create a new variable that had the
> trimmed results ("char *cp = line;"), after that commit it would just trim
> the line itself. Which is correct for the case of it being a header, but
> if it's the first non-header line, it's wrong.
True; trimming the body is obviously wrong.
But when is it correct to ltrim a header line? It means we are going to
accept a header (or header-looking line in body) that is indented. I
don't know why 87ab799 (builtin-mailinfo.c 2007-03-12) was coded that way.
builtin-mailinfo.c | 3 +--
t/t5100/msg0015 | 2 +-
2 files changed, 2 insertions(+), 3 deletions(-)
diff --git a/builtin-mailinfo.c b/builtin-mailinfo.c
index a50ac22..ce2ef6b 100644
--- a/builtin-mailinfo.c
+++ b/builtin-mailinfo.c
@@ -779,8 +779,7 @@ static int handle_commit_msg(struct strbuf *line)
return 0;
if (still_looking) {
- strbuf_ltrim(line);
- if (!line->len)
+ if (!line->len || (line->len == 1 && line->buf[0] == '\n'))
return 0;
}
diff --git a/t/t5100/msg0015 b/t/t5100/msg0015
index 9577238..4abb3d5 100644
--- a/t/t5100/msg0015
+++ b/t/t5100/msg0015
@@ -1,2 +1,2 @@
-- a list
+ - a list
- of stuff
--
To unsubscribe from this list: send the line "unsubscribe git" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html