Re: [PATCH v2] t5411: consistent result for proc-receive broken test
- From
Jiang Xin <worldhello.net@gmail.com>
- Date
- Nov 10, 2020, 11:44 UTC
- Message-ID
- <CANYiYbEZu9Oq-X=o9aVzNz7BU=8HevnPCuGk=8VQSsz-o2BO6Q@mail.gmail.com>
- In-Reply-To
- <xmqqh7pyb61f.fsf@gitster.c.googlers.com>
Junio C Hamano <gitster@pobox.com> 于2020年11月10日周二 上午4:59写道：
Show 11 quoted lines
>
> > @@ -166,6 +177,8 @@ int cmd__proc_receive(int argc, const char **argv)
> > fprintf(stderr, "proc-receive> %s\n", item->string);
> > }
> >
> > + if (die_report)
> > + die("die with the --die-report option");
>
> And at this point we have already read everything the other end
> said (if so, there is no need for the artificial "read everything
> before we die")?In patch v3, will use "gently" forms of packet_write_fmt() and packet_flush(), and it is no necessary to read everything before die. And will add more "--die-*" option in test helper.
Show 11 quoted lines
> > diff --git a/t/t5411/test-0013-bad-protocol.sh b/t/t5411/test-0013-bad-protocol.sh > > index c5fe4cb37b..5c5241bc95 100644 > > --- a/t/t5411/test-0013-bad-protocol.sh > > +++ b/t/t5411/test-0013-bad-protocol.sh > > @@ -55,19 +55,16 @@ test_expect_success "proc-receive: bad protocol (hook --die-version, $PROTOCOL)" > > test_must_fail git -C workbench push origin \ > > HEAD:refs/for/master/topic \ > > >out 2>&1 && > > Are these expected to conflict with Dscho's changes to move 'master' > around?
See topic "gitster/js/default-branch-name-adjust-t5411" has been merge to master branch already. Will rebase on it.
Show 44 quoted lines
> > - make_user_friendly_and_stable_output <out >actual &&
> > -
> > + make_user_friendly_and_stable_output <out |
> > + sed -n \
> > + -e "/^To / { s/ */ /g; p; }" \
> > + -e "/^ ! / { s/ */ /g; p; }" \
> > + >actual &&
>
> It's the same thing but I somehow find "s/ */ /g" easier to read.
> The comparison is between "there may be two things or more---squish
> them down to one" and "After one thing, there may be any number of
> things---remove all the extra ones".
>
> Makes me wonder if make_user_friendly should optionally have an
> option to do something like this for us. I doubt it that it is
> worth to do something like the attached patch.
>
>
> t/t5411/common-functions.sh | 7 +++++--
> 1 file changed, 5 insertions(+), 2 deletions(-)
>
> diff --git c/t/t5411/common-functions.sh w/t/t5411/common-functions.sh
> index 6580bebd8e..6919639c60 100644
> --- c/t/t5411/common-functions.sh
> +++ w/t/t5411/common-functions.sh
> @@ -40,7 +40,9 @@ create_commits_in () {
> # `GIT_TEST_GETTEXT_POISON=true` in order to test unintentional translations
> # on plumbing commands.
> make_user_friendly_and_stable_output () {
> - sed \
> + local en=
> + case "$#" in 0) ;; *) en=-n ;; esac
> + sed $en \
> -e "s/ *\$//" \
> -e "s/ */ /g" \
> -e "s/'/\"/g" \
> @@ -52,5 +54,6 @@ make_user_friendly_and_stable_output () {
> -e "s/$(echo $A | cut -c1-7)[0-9a-f]*/<OID-A>/g" \
> -e "s/$(echo $B | cut -c1-7)[0-9a-f]*/<OID-B>/g" \
> -e "s#To $URL_PREFIX/upstream.git#To <URL/of/upstream.git>#" \
> - -e "/^error: / d"
> + -e "/^error: / d" \
> + ${1+"$@"}
> }If we call make_user_friendly_and_stable_output like this:
make_user_friendly_and_stable_output \
-e "/^To / { p; n; p; n; p; }Text lines next to "^To " will not be formatted by the built-in sed's arguments. So will write like this:
-- snip --
diff --git a/t/t5411/common-functions.sh b/t/t5411/common-functions.sh index 521a347710..b7cca2d8fb 100644 --- a/t/t5411/common-functions.sh +++ b/t/t5411/common-functions.sh @@ -42,7 +42,7 @@ create_commits_in () { make_user_friendly_and_stable_output () { sed \ -e "s/ *\$//" \ - -e "s/ */ /g" \ + -e "s/ */ /g" \ -e "s/'/\"/g" \ -e "s/ / /g" \ -e "s/$A/<COMMIT-A>/g" \ @@ -52,5 +52,11 @@ make_user_friendly_and_stable_output () { -e "s/$(echo $A | cut -c1-7)[0-9a-f]*/<OID-A>/g" \ -e "s/$(echo $B | cut -c1-7)[0-9a-f]*/<OID-B>/g" \ -e "s#To $URL_PREFIX/upstream.git#To <URL/of/upstream.git>#" \ - -e "/^error: / d" + -e "/^error: / d" | \ + if test $# -eq 0 + then + cat + else + sed ${1+"$@"} + fi } -- snap --