Re: [PATCH 08/12] t5520: use test_cmp_rev where possible
- From
Denton Liu <liu.denton@gmail.com>
- Date
- Oct 18, 2019, 18:52 UTC
- Message-ID
- <20191018185227.GA74877@generichostname>
- In-Reply-To
- <CAPig+cRF0taNhmZVu0z+P+Oe3em11Jz33GY8eBZRTGkx3z=TOw@mail.gmail.com>
Hi Eric,
Thanks for the reviews. I have no idea how you always get to my patches so quickly but I appreciate the prompt reviews whenever I send a test-related patchset in.
I've fixed up the other concerns you had and I'll send a v2 later but I wanted to address this one.
On Thu, Oct 17, 2019 at 07:41:44PM -0400, Eric Sunshine wrote:
Show 20 quoted lines
> On Thu, Oct 17, 2019 at 7:17 PM Denton Liu <liu.denton@gmail.com> wrote: > > In case an invocation of `git rev-list` fails within the subshell, the > > failure will be masked. Remove the subshell and use test_cmp_rev() so > > that failures can be discovered. > > > > Signed-off-by: Denton Liu <liu.denton@gmail.com> > > --- > > diff --git a/t/t5520-pull.sh b/t/t5520-pull.sh > > @@ -597,10 +597,10 @@ test_expect_success 'pull --rebase dies early with dirty working directory' ' > > test_must_fail git pull && > > - test "$COPY" = "$(git rev-parse --verify me/copy)" && > > + test_cmp_rev "$COPY" me/copy && > > This transformation doesn't look correct. COPY already holds the > result of a git-rev-parse invocation: > > COPY="$(git rev-parse --verify me/copy)" && > > so passing it to test_cmp_rev() -- which applies its own git-rev-parse > invocation -- doesn't make sense.
I'll annotate the entire test case with my comments:
# so we'll be testing that pull --rebase dies early test_expect_success 'pull --rebase dies early with dirty working directory' ' git checkout to-rebase && git update-ref refs/remotes/me/copy copy^ && COPY="$(git rev-parse --verify me/copy)" && git rebase --onto $COPY copy &&
# according to git log --graph, we have this currently: # # $ git log --pretty=oneline --graph to-rebase copy me/copy # * 9366795adb50ef6eb482b610b37cb1fb6edbd3d0 (HEAD -> to-rebase) to-rebase # * b3dcf50eea47c4ad85faabb0fb74eded71cc829f file3 # * faf459539411b4557cf735232a3746be073177a9 new file # | * f340b1b8932f7b1e016c06867cbdc3f637eeea2d (copy) conflict # |/ # * 5e86d50a28977ebee5ea378f81591ea558149272 (me/second, me/copy, second) modified # * d25cf184567afad1dc1e018b2b7d793bc1bd2dc1 original
test_config branch.to-rebase.remote me && test_config branch.to-rebase.merge refs/heads/copy && test_config branch.to-rebase.rebase true &&
# we make the tree dirty here echo dirty >>file && git add file &&
# and over here, the pull should fail test_must_fail git pull &&
# but since we're testing that it dies early, we want to # make sure that the remote ref doesn't change, which is # why it should still be equal test_cmp_rev "$COPY" me/copy &&
git checkout HEAD -- file &&
# but over here, the pull succeeds... git pull &&
# so as a result, the remote ref should now be updated test_cmp_rev ! "$COPY" me/copy '
So after grokking the test case, it seems like the the transformation is indeed correct. Maybe we can replace the last line with
test_cmp_rev copy me/copy
but I think I'll leave it unless you have any strong opinions.
Thanks,
Denton