On Tue, Jun 01 2021, Jiang Xin wrote:
> From: Jiang Xin <zhiyou.jx@xxxxxxxxxxxxxxx>
>
> We used to call `make_user_friendly_and_stable_output` to mangle
> trailing spaces in output before comparing with the expect file.
> Ævar recommends generating expect file using pattern "'s/Z$//'" to
> compare expect file with raw output.
I've tested both of these patches and they fix the reported issue on
that gcc45 machine.
> Suggested-by: Ævar Arnfjörð Bjarmason <avarab@xxxxxxxxx>
> Signed-off-by: Jiang Xin <zhiyou.jx@xxxxxxxxxxxxxxx>
> ---
> t/t6020-bundle-misc.sh | 89 ++++++++++++++++++++++--------------------
> 1 file changed, 46 insertions(+), 43 deletions(-)
>
> diff --git a/t/t6020-bundle-misc.sh b/t/t6020-bundle-misc.sh
> index c6a8ea7f76..7ee43fb1df 100755
> --- a/t/t6020-bundle-misc.sh
> +++ b/t/t6020-bundle-misc.sh
> @@ -83,7 +83,7 @@ test_commit_setvar () {
>
> # Format the output of git commands to make a user-friendly and stable
> # text. We can easily prepare the expect text without having to worry
> -# about future changes of the commit ID and spaces of the output.
> +# about future changes of the commit ID.
> make_user_friendly_and_stable_output () {
> sed \
> -e "s/$A[0-9a-f]*/<COMMIT-A>/g" \
> @@ -104,8 +104,11 @@ make_user_friendly_and_stable_output () {
> -e "s/$P[0-9a-f]*/<COMMIT-P>/g" \
> -e "s/$TAG1[0-9a-f]*/<TAG-1>/g" \
> -e "s/$TAG2[0-9a-f]*/<TAG-2>/g" \
> - -e "s/$TAG3[0-9a-f]*/<TAG-3>/g" \
> - -e "s/ *\$//"
> + -e "s/$TAG3[0-9a-f]*/<TAG-3>/g"
> +}
> +
> +format_and_save_expect () {
> + sed -e 's/^> //' -e 's/Z$//' >expect
> }
>
> # (C) (D, pull/1/head, topic/1)
> @@ -180,11 +183,11 @@ test_expect_success 'create bundle from special rev: main^!' '
>
> git bundle verify special-rev.bdl |
> make_user_friendly_and_stable_output >actual &&
> - cat >expect <<-\EOF &&
> - The bundle contains this ref:
> - <COMMIT-P> refs/heads/main
> - The bundle requires this ref:
> - <COMMIT-O>
> + format_and_save_expect <<-\EOF &&
> + > The bundle contains this ref:
> + > <COMMIT-P> refs/heads/main
> + > The bundle requires this ref:
> + > <COMMIT-O> Z
> EOF
> test_cmp expect actual &&
I think for this test it would be better to just have the "Z" suffix and
leave off prefixing all the lines with ">". That's done in t0000.sh
because we have various empty lines, but here that's not the case.
I don't think that nit is worth a re-roll, and as you noted you're doing
some larger changes to these tests in follow-up patches.
It's just convenient to have the test_cmp expect input as close to
copy/paste-able as possible.
