Re: [PATCH 1/6] t/send-email.sh: add test for suppress self

On Wed, May 29, 2013 at 01:28:52PM -0700, Junio C Hamano wrote:

"Michael S. Tsirkin" [off-list ref] writes:

quoted

Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
---

Thanks.

Thanks, I'll address your comments and repost.
You asked some questions below, so I tried to answer.

quoted

 t/t9001-send-email.sh | 41 +++++++++++++++++++++++++++++++++++++++++
 1 file changed, 41 insertions(+)

diff --git a/t/t9001-send-email.sh b/t/t9001-send-email.sh
index ebd5c5d..36ecf73 100755
--- a/t/t9001-send-email.sh
+++ b/t/t9001-send-email.sh

@@ -171,6 +171,47 @@ Result: OK
 EOF
 "
 
+test_suppress_self () {
+		test_commit $3 &&
+		test_when_finished "git reset --hard HEAD^" &&
+		{
+			echo "#!$SHELL_PATH"
+			echo sed -n -e s/^cccmd--//p \"\$1\"
+		} > cccmd-sed &&
+		chmod +x cccmd-sed &&

We can use write_script for this kind of thing, I think.

Important?
It's open-coded elsewhere in this test.

quoted

+		git commit --amend --author="$1 <$2>" -F - << EOF && \

Hmm,...  everything below this function is fed as the standard input
to "git commit" as its updated log message (i.e. "--amend -F -")?

Puzzled...
The EOF I can find is at the very bottom of this function, so there
is no "next command" that && at the end of the above line is
cascading the control to.

Doubly puzzled...

In any case, please do not add " \" at the end of line when the line
ends one command and "&&" at the end of line clearly tells the shell
that you haven't stopped talking yet.

Note that \ make following lines count as continuation of this one.
So they are *not* part of standard input.

Look here:

FOO << EOF && BAR && VAR
BLABLA
EOF

this feeds BLABLA as input to FOO, then
runs BAR and then VAR.

Now we don't want to put BAR and VAR on same line
because that line is too long.
So we write the equivalent: \ followed by newline
is same as space for shell, so we can write it as:

FOO << EOF && \
 BAR && \
 VAR
BLABLA
EOF

Clear now?

If we don't want to use \, this can also
be done like this:

FOO << EOF && 
BLABLA
EOF
 BAR && 
 VAR

I think this is what you suggest.

quoted

+		clean_fake_sendmail && \
+		echo suppress-self-$3.patch > /dev/tty && \

Do we always have /dev/tty?  If this is a leftover debugging, please
remove it.

Leftover.

 If redirecting it to >&2 does not upset what the test
does, that is good, too (you can run the test with -v option to view
the output).

quoted

+		git format-patch --stdout -1 >suppress-self-$3.patch && \
+		git send-email --from="$1 <$2>" \
+		--to=nobody@example.com \
+		--cc-cmd=./cccmd-sed \
+		--suppress-cc=self \
+		--smtp-server="$(pwd)/fake.sendmail" \
+		suppress-self-$3.patch && \
+		mv msgtxt1 msgtxt1-$3 && \
+		sed -e '/^$/q' msgtxt1-$3 > msghdr1-$3 && \

Style.  No SP between redirection operator and redirection target,
e.g. >$filename, <$filename, etc.  Some in this patch is done
correctly (e.g. format-patch above), some others are not.

will fix.

quoted

+		rm -f expected-no-cc-$3 && \
+		touch expected-no-cc-$3 && \

Please reserve "touch" for "make sure it has recent timestamp", not
for "make sure it exists and is empty".  The above two should be
more like:

		>"expected-no-cc-$3" &&

OK

Also, even though it is not required by POSIX, please dquote the
redirection target filename if you have variable expansion.  Some
versions of bash (incorrectly) give warning if you don't.

OK

quoted

+		grep '^Cc:' msghdr1-$3 > actual-no-cc-$3 && \
+		test_cmp expected-no-cc-$3 actual-no-cc-$3
+test suppress-cc.self $3 with name $1 email $2
+
+$3
+
+cccmd--"$1" <$2>
+
+Cc: "$1" <$2>
+Cc: $1 <$2>
+Signed-off-by: "$1" <$2>
+Signed-off-by: $1 <$2>
+EOF
+}
+
+test_expect_success $PREREQ 'self name is suppressed' "
+	test_suppress_self 'A U Thor' 'author@redhat.com' 'self_name_suppressed'
+"
+
 test_expect_success $PREREQ 'Show all headers' '
 	git send-email \
 		--dry-run \