On Dec 19, 2016, at 15:26, Junio C Hamano wrote:
> "Kyle J. McKay" <mack...@gmail.com> writes:
>
>>> OK. So we do not expect it to fail, but we still do want the side
>>> effect of that function (i.e. accmulation into the field).
>>>
>>> Somebody care to send a final "agreed-upon" version?
>>
>> Yup, here it is:
>
> Thanks.
Whoops. there's an extra paragraph in the commit description that I
meant to remove and, of course, I didn't notice it until I sent the
copy to the list. :(
I don't think a "fixup" or "squash" can replace a description, right?
So here's a replacement patch with the correct description with the
deleted paragrah:
-- >8 --
Since 6b4b013f18 (mailinfo: handle in-body header continuations,
2016-09-20, v2.11.0) mailinfo.c has contained new code with an
assert of the form:
assert(call_a_function(...))
The function in question, check_header, has side effects. This
means that when NDEBUG is defined during a release build the
function call is omitted entirely, the side effects do not
take place and tests (fortunately) start failing.
Since the only time that mi->inbody_header_accum is appended to is
in check_inbody_header, and appending onto a blank
mi->inbody_header_accum always happens when is_inbody_header is
true, this guarantees a prefix that causes check_header to always
return true.
Therefore replace the assert with an if !check_header + DIE
combination to reflect this.
Helped-by: Jonathan Tan <jonathanta...@google.com>
Helped-by: Jeff King <p...@peff.net>
Acked-by: Johannes Schindelin <johannes.schinde...@gmx.de>
Signed-off-by: Kyle J. McKay <mack...@gmail.com>
---
Notes:
Please include this PATCH in 2.11.x maint
mailinfo.c | 3 ++-
1 file changed, 2 insertions(+), 1 deletion(-)
diff --git a/mailinfo.c b/mailinfo.c
index 2fb3877e..a489d9d0 100644
--- a/mailinfo.c
+++ b/mailinfo.c
@@ -710,7 +710,8 @@ static void flush_inbody_header_accum(struct mailinfo *mi)
{
if (!mi->inbody_header_accum.len)
return;
- assert(check_header(mi, &mi->inbody_header_accum, mi->s_hdr_data, 0));
+ if (!check_header(mi, &mi->inbody_header_accum, mi->s_hdr_data, 0))
+ die("BUG: inbody_header_accum, if not empty, must always
contain a valid in-body header");
strbuf_reset(&mi->inbody_header_accum);
}
---
