# git diff --submodule=diff fails with submodules in a submodule 5 messages from 2017-03-31 to 2017-03-31. Participants: David Parrish, Stefan Beller, Jacob Keller, Junio C Hamano. Thread: https://gitlist.dev/t/45566 ## David Parrish, 2017-03-31 15:40 Subject: git diff --submodule=diff fails with submodules in a submodule Message-ID: URL: https://gitlist.dev/e/CAPd5SoV1F7MMDSRGzrd-7soty%3D61vnumQ02UtBJs2fwEyv%2BV3Q%40mail.gmail.com ``` When I try to run `git diff --submodule=diff` in a submodule which has it's own submodules that have changes I get the error: fatal: bad object Let me know if you need an example reproduce the issue. David ``` ## Stefan Beller, 2017-03-31 17:07 Subject: Re: git diff --submodule=diff fails with submodules in a submodule Message-ID: URL: https://gitlist.dev/e/CAGZ79kapamr4%3DC_JQ4Ud2bF6NoXn8qeQR1fYScWB4%2B-EzEckmQ%40mail.gmail.com In-Reply-To: ``` +cc Jacob, who implemented --submodule=diff On Fri, Mar 31, 2017 at 8:40 AM, David Parrish wrote: > When I try to run `git diff --submodule=diff` in a submodule which has > it's own submodules that have changes I get the error: fatal: bad > object Thanks for the bug report! > Let me know if you need an example reproduce the issue. I could reproduce it when playing around locally with a submodule in submodules. I think sub-submodule needs to have its HEAD moved from the recorded commit. Thanks, Stefan ``` ## Jacob Keller, 2017-03-31 17:31 Subject: Re: git diff --submodule=diff fails with submodules in a submodule Message-ID: URL: https://gitlist.dev/e/CA%2BP7%2BxoqWCpySc17104zggLgwx1mc0T%2BJXybrgd8sQxoQFP-PQ%40mail.gmail.com In-Reply-To: ``` On Fri, Mar 31, 2017 at 10:07 AM, Stefan Beller wrote: > +cc Jacob, who implemented --submodule=diff > > On Fri, Mar 31, 2017 at 8:40 AM, David Parrish wrote: >> When I try to run `git diff --submodule=diff` in a submodule which has >> it's own submodules that have changes I get the error: fatal: bad >> object > > Thanks for the bug report! > >> Let me know if you need an example reproduce the issue. > > I could reproduce it when playing around locally with a submodule in > submodules. I think sub-submodule needs to have its HEAD moved from > the recorded commit. > > Thanks, > Stefan Hmm. An example reproduction would be helpful. Ideally in the form of a test ;) But otherwise whatever helps. I will try to look at this, but I'm busy for a few days. Thanks, Jake ``` ## Stefan Beller, 2017-03-31 17:59 Subject: [PATCH] failure with diff --submodule=diff with moved nested submodule HEAD Message-ID: <20170331175907.17400-1-sbeller@google.com> URL: https://gitlist.dev/e/20170331175907.17400-1-sbeller%40google.com In-Reply-To: ``` This fails reliable for me. Signed-off-by: Stefan Beller --- t/t4060-diff-submodule-option-diff-format.sh | 16 ++++++++++++++++ 1 file changed, 16 insertions(+) diff --git a/t/t4060-diff-submodule-option-diff-format.sh b/t/t4060-diff-submodule-option-diff-format.sh index 7e23b55ea4..89bced3484 100755 --- a/t/t4060-diff-submodule-option-diff-format.sh +++ b/t/t4060-diff-submodule-option-diff-format.sh @@ -746,4 +746,20 @@ test_expect_success 'diff --submodule=diff with .git file' ' test_cmp expected actual ' +test_expect_success 'setup nested submodule' ' + git submodule add -f ./sm2 && + git commit -a -m "add sm2" && + git -C sm2 submodule add ../sm2 && + git -C sm2 commit -a -m "nested sub" +' + +test_expect_success 'move nested submodule HEAD' ' + git -C sm2/sm2 commit --allow-empty -m "new HEAD" +' + +test_expect_success 'diff --submodule=diff with moved nested submodule HEAD' ' + git -C sm2 diff --submodule=diff >actual 2>err && + test_must_be_empty err +' + test_done -- 2.12.2.511.g2abb8caf66 ``` ## Junio C Hamano, 2017-03-31 19:38 Subject: Re: [PATCH] failure with diff --submodule=diff with moved nested submodule HEAD Message-ID: URL: https://gitlist.dev/e/xmqqk275xmmu.fsf%40gitster.mtv.corp.google.com In-Reply-To: <20170331175907.17400-1-sbeller@google.com> ``` Stefan Beller writes: > This fails reliable for me. > > Signed-off-by: Stefan Beller > --- I take it that this is a WIP for the testing half of the patch you are working on (i.e. you are not just throwing a patch to document known breakage to be fixed later, in which case these would have been _expect_failure). Thanks. > t/t4060-diff-submodule-option-diff-format.sh | 16 ++++++++++++++++ > 1 file changed, 16 insertions(+) > > diff --git a/t/t4060-diff-submodule-option-diff-format.sh b/t/t4060-diff-submodule-option-diff-format.sh > index 7e23b55ea4..89bced3484 100755 > --- a/t/t4060-diff-submodule-option-diff-format.sh > +++ b/t/t4060-diff-submodule-option-diff-format.sh > @@ -746,4 +746,20 @@ test_expect_success 'diff --submodule=diff with .git file' ' > test_cmp expected actual > ' > > +test_expect_success 'setup nested submodule' ' > + git submodule add -f ./sm2 && > + git commit -a -m "add sm2" && > + git -C sm2 submodule add ../sm2 && > + git -C sm2 commit -a -m "nested sub" > +' > + > +test_expect_success 'move nested submodule HEAD' ' > + git -C sm2/sm2 commit --allow-empty -m "new HEAD" > +' > + > +test_expect_success 'diff --submodule=diff with moved nested submodule HEAD' ' > + git -C sm2 diff --submodule=diff >actual 2>err && > + test_must_be_empty err > +' > + > test_done ```