{"thread":{"id":"10938","subject":"t9106 failure, bisect weirdness","startedAt":"2007-11-19T23:06:01Z","lastAt":"2007-11-21T09:57:27Z","messageCount":7,"participants":["carbonated beverage","Christian Couder","Eric Wong"],"isPatch":false,"patchVersion":null,"patchTotal":null},"messages":[{"id":"60379","messageId":"20071119230601.GA15624@net-ronin.org","threadId":"10938","inReplyTo":null,"subject":"t9106 failure, bisect weirdness","fromName":"carbonated beverage","fromEmail":"ramune@net-ronin.org","sentAt":"2007-11-19T23:06:01Z","receivedAt":"2007-11-19T23:06:01Z","isPatch":false,"sender":{"key":"ramune@net-ronin.org","avatar":null},"body":"Hi all,\n\nJust checked out v1.5.3.6 from git, built it on a Debian/stable i386 box,\nand got a failure in one of the test suites:\n\n*** t9106-git-svn-dcommit-clobber-series.sh ***\n* ok 1: initialize repo\n* ok 2: (supposedly) non-conflicting change from SVN\n* ok 3: some unrelated changes to git\n* FAIL 4: change file but in unrelated area\n\n test x\"`sed -n -e 4p < file`\" = x4 &&\n test x\"`sed -n -e 7p < file`\" = x7 &&\n perl -i -p -e 's/^4$/4444/' file &&\n perl -i -p -e 's/^7$/7777/' file &&\n test x\"`sed -n -e 4p < file`\" = x4444 &&\n test x\"`sed -n -e 7p < file`\" = x7777 &&\n git commit -m '4 => 4444, 7 => 7777' file &&\n git svn dcommit &&\n svn up tmp &&\n cd tmp &&\n test x\"`sed -n -e 4p < file`\" = x4444 &&\n test x\"`sed -n -e 7p < file`\" = x7777 &&\n test x\"`sed -n -e 58p < file`\" = x5588 &&\n test x\"`sed -n -e 61p < file`\" = x6611\n\n* ok 5: attempt to dcommit with a dirty index\n* failed 1 among 5 test(s)\nmake[1]: *** [t9106-git-svn-dcommit-clobber-series.sh] Error 1\nmake[1]: Leaving directory `/home/ramune/src/git/git/t'\nmake: *** [test] Error 2\nramune/lycaeum:git:\n\nBisecting gives me:\n\nfb159580a1628947f0a088e24cfe6fe4c81d99d0 is first bad commit\ncommit fb159580a1628947f0a088e24cfe6fe4c81d99d0\nAuthor: Eric Wong \nDate: Mon Nov 5 03:21:48 2007 -0800\n\n git-svn: t9114: verify merge commit message in test\n\n It's possible that we end up with an incorrect commit message\n in this test after making changes to fix the clobber bug\n in dcommit.\n\n Signed-off-by: Eric Wong \n Signed-off-by: Junio C Hamano \n\n:040000 040000 1c4f3776c031f9531869e736b661a0559fab5ff6 4f4c5ecd6c79f7e9732b554e2f8dd536a8a8f2a6 M t\n\nWhich seems odd -- the previous commit (marked good, because make test didn't\nfail) was the one that modified t9106-git-svn-dcommit-clobber-series.sh.\n\nI tried running the test manually and noticed something weird as well:\n\nramune/lycaeum:t: sh t9106-git-svn-dcommit-clobber-series.sh\n* ok 1: initialize repo\n* ok 2: (supposedly) non-conflicting change from SVN\n* ok 3: some unrelated changes to git\n* ok 4: change file but in unrelated area\n* passed all 4 test(s)\nramune/lycaeum:t:\nramune/lycaeum:t: pwd\n/home/ramune/src/git/git/t\nramune/lycaeum:t:\nramune/lycaeum:t: sh t9106-git-svn-dcommit-clobber-series.sh\n* ok 1: initialize repo\n* ok 2: (supposedly) non-conflicting change from SVN\n* ok 3: some unrelated changes to git\n* FAIL 4: change file but in unrelated area\n\n test x\"`sed -n -e 4p < file`\" = x4 &&\n test x\"`sed -n -e 7p < file`\" = x7 &&\n perl -i -p -e 's/^4$/4444/' file &&\n perl -i -p -e 's/^7$/7777/' file &&\n test x\"`sed -n -e 4p < file`\" = x4444 &&\n test x\"`sed -n -e 7p < file`\" = x7777 &&\n git commit -m '4 => 4444, 7 => 7777' file &&\n git svn dcommit &&\n svn up tmp &&\n cd tmp &&\n test x\"`sed -n -e 4p < file`\" = x4444 &&\n test x\"`sed -n -e 7p < file`\" = x7777 &&\n test x\"`sed -n -e 58p < file`\" = x5588 &&\n test x\"`sed -n -e 61p < file`\" = x6611\n\n* failed 1 among 4 test(s)\n\nSo it succeeds once, then fails.\n\n/bin/sh is dash 0.5.3-7 from Debian.\n\nAny additional information needed?\n\n-- DN\nDaniel\n"},{"id":"60387","messageId":"200711200552.27001.chriscool@tuxfamily.org","threadId":"10938","inReplyTo":"20071119230601.GA15624@net-ronin.org","subject":"Re: t9106 failure, bisect weirdness","fromName":"Christian Couder","fromEmail":"chriscool@tuxfamily.org","sentAt":"2007-11-20T04:52:26Z","receivedAt":"2007-11-20T04:52:26Z","isPatch":false,"sender":{"key":"christian.couder@gmail.com","avatar":"https://avatars.githubusercontent.com/u/208954?v=4"},"body":"Le mardi 20 novembre 2007, carbonated beverage a écrit :\n[...]\n> I tried running the test manually and noticed something weird as well:\n>\n> ramune/lycaeum:t: sh t9106-git-svn-dcommit-clobber-series.sh\n> * ok 1: initialize repo\n> * ok 2: (supposedly) non-conflicting change from SVN\n> * ok 3: some unrelated changes to git\n> * ok 4: change file but in unrelated area\n> * passed all 4 test(s)\n> ramune/lycaeum:t:\n> ramune/lycaeum:t: pwd\n> /home/ramune/src/git/git/t\n> ramune/lycaeum:t:\n> ramune/lycaeum:t: sh t9106-git-svn-dcommit-clobber-series.sh\n> * ok 1: initialize repo\n> * ok 2: (supposedly) non-conflicting change from SVN\n> * ok 3: some unrelated changes to git\n> * FAIL 4: change file but in unrelated area\n>\n> test x\"`sed -n -e 4p < file`\" = x4 &&\n> test x\"`sed -n -e 7p < file`\" = x7 &&\n> perl -i -p -e 's/^4$/4444/' file &&\n> perl -i -p -e 's/^7$/7777/' file &&\n> test x\"`sed -n -e 4p < file`\" = x4444 &&\n> test x\"`sed -n -e 7p < file`\" = x7777 &&\n> git commit -m '4 => 4444, 7 => 7777' file &&\n> git svn dcommit &&\n> svn up tmp &&\n> cd tmp &&\n> test x\"`sed -n -e 4p < file`\" = x4444 &&\n> test x\"`sed -n -e 7p < file`\" = x7777 &&\n> test x\"`sed -n -e 58p < file`\" = x5588 &&\n> test x\"`sed -n -e 61p < file`\" = x6611\n>\n> * failed 1 among 4 test(s)\n>\n> So it succeeds once, then fails.\n>\n> /bin/sh is dash 0.5.3-7 from Debian.\n>\n> Any additional information needed?\n\nCould you try running the test twice when you bisect ?\nThis way bisecting should give you the real first bad commit. \n\nCould you also try to find the line that fails in the 4th test ?\nFor example you could replace all the trailing \"&&\" in this test with a new \nline that contains \"echo $?\".\n\nThanks in advance,\nChristian.\n"},{"id":"60391","messageId":"20071120061314.GA21819@net-ronin.org","threadId":"10938","inReplyTo":"200711200552.27001.chriscool@tuxfamily.org","subject":"Re: t9106 failure, bisect weirdness","fromName":"carbonated beverage","fromEmail":"ramune@net-ronin.org","sentAt":"2007-11-20T06:13:14Z","receivedAt":"2007-11-20T06:13:14Z","isPatch":false,"sender":{"key":"ramune@net-ronin.org","avatar":null},"body":"> Could you try running the test twice when you bisect ?\n> This way bisecting should give you the real first bad commit. \n\nI re-ran it several times, and also manually -- there were times it'd\nsucceed 6 times in a row before failing. Weird.\n\nFinally, the bisect came down to:\n\n ramune/lycaeum:git: git bisect good\n c74d9acf20ba0c69bbd67c5b0bb3bd3c2349cebe is first bad commit\n commit c74d9acf20ba0c69bbd67c5b0bb3bd3c2349cebe\n Author: Eric Wong \n Date: Mon Nov 5 03:21:47 2007 -0800\n \n git-svn: fix dcommit clobbering when committing a series of diffs\n\nModifying the failing t9106-git-svn-dcommit-clobber-series.sh test to\nadd touch statements as so:\n\n test x\\\"\\`sed -n -e 4p < file\\`\\\" = x4444 && touch '/tmp/eleven' &&\n test x\\\"\\`sed -n -e 7p < file\\`\\\" = x7777 && touch '/tmp/twelve' &&\n test x\\\"\\`sed -n -e 58p < file\\`\\\" = x5588 && touch '/tmp/thirteen' &&\n test x\\\"\\`sed -n -e 61p < file\\`\\\" = x6611\n\nLine 52 was the culprit:\n\n test x\\\"\\`sed -n -e 58p < file\\`\\\" = x5588 &&\n\n-- DN\nDaniel\n"},{"id":"60452","messageId":"200711210508.27455.chriscool@tuxfamily.org","threadId":"10938","inReplyTo":"20071120061314.GA21819@net-ronin.org","subject":"Re: t9106 failure, bisect weirdness","fromName":"Christian Couder","fromEmail":"chriscool@tuxfamily.org","sentAt":"2007-11-21T04:08:27Z","receivedAt":"2007-11-21T04:08:27Z","isPatch":false,"sender":{"key":"christian.couder@gmail.com","avatar":"https://avatars.githubusercontent.com/u/208954?v=4"},"body":"Le mardi 20 novembre 2007, carbonated beverage a écrit :\n> > Could you try running the test twice when you bisect ?\n> > This way bisecting should give you the real first bad commit.\n>\n> I re-ran it several times, and also manually -- there were times it'd\n> succeed 6 times in a row before failing. Weird.\n>\n> Finally, the bisect came down to:\n>\n> ramune/lycaeum:git: git bisect good\n> c74d9acf20ba0c69bbd67c5b0bb3bd3c2349cebe is first bad commit\n> commit c74d9acf20ba0c69bbd67c5b0bb3bd3c2349cebe\n> Author: Eric Wong \n> Date: Mon Nov 5 03:21:47 2007 -0800\n>\n> git-svn: fix dcommit clobbering when committing a series of diffs\n\nThanks for having done that. Unfortunately I don't know this part of the \ncode, so I cc'ed Eric.\n\n> Modifying the failing t9106-git-svn-dcommit-clobber-series.sh test to\n> add touch statements as so:\n>\n> test x\\\"\\`sed -n -e 4p < file\\`\\\" = x4444 && touch '/tmp/eleven' &&\n> test x\\\"\\`sed -n -e 7p < file\\`\\\" = x7777 && touch '/tmp/twelve' &&\n> test x\\\"\\`sed -n -e 58p < file\\`\\\" = x5588 && touch '/tmp/thirteen' &&\n> test x\\\"\\`sed -n -e 61p < file\\`\\\" = x6611\n>\n> Line 52 was the culprit:\n>\n> test x\\\"\\`sed -n -e 58p < file\\`\\\" = x5588 &&\n\nOk thanks for doing that too.\nCould you also look at the \"file\" when the test succeed and when it does not \nand send us both versions and a diff between them (if it's not too big).\n\nThanks in advance,\nChristian.\n"},{"id":"60455","messageId":"20071121045638.GA9184@net-ronin.org","threadId":"10938","inReplyTo":"200711210508.27455.chriscool@tuxfamily.org","subject":"Re: t9106 failure, bisect weirdness","fromName":"carbonated beverage","fromEmail":"ramune@net-ronin.org","sentAt":"2007-11-21T04:56:38Z","receivedAt":"2007-11-21T04:56:38Z","isPatch":false,"sender":{"key":"ramune@net-ronin.org","avatar":null},"body":"> Ok thanks for doing that too.\n> Could you also look at the \"file\" when the test succeed and when it does not \n> and send us both versions and a diff between them (if it's not too big).\n\nSize-wise, they're tiny -- though there's a lot of lines.\n\nThe diff:\n\nramune/lycaeum:t: diff -u works.t fails.t \n--- works.t 2007-11-20 21:54:29.000000000 -0700\n+++ fails.t 2007-11-20 21:54:39.000000000 -0700\n@@ -55,9 +55,9 @@\n 55\n 56\n 57\n-5588\n+58\n 59\n 60\n-6611\n+61\n 62\n 63\n\nAnd the full (working) file:\n\n1\n2\n3\n4444\n5\n6\n7777\n8\n9\n10\n11\n12\n13\n14\n15\n16\n17\n18\n19\n20\n21\n22\n23\n24\n25\n26\n27\n28\n29\n30\n31\n32\n33\n34\n35\n36\n37\n38\n39\n40\n41\n42\n43\n44\n45\n46\n47\n48\n49\n50\n51\n52\n53\n54\n55\n56\n57\n5588\n59\n60\n6611\n62\n63\n"},{"id":"60483","messageId":"20071121091001.GA23266@soma","threadId":"10938","inReplyTo":"20071121045638.GA9184@net-ronin.org","subject":"Re: t9106 failure, bisect weirdness","fromName":"Eric Wong","fromEmail":"normalperson@yhbt.net","sentAt":"2007-11-21T09:10:01Z","receivedAt":"2007-11-21T09:10:01Z","isPatch":false,"sender":{"key":"e@80x24.org","avatar":null},"body":"Hi, thanks for the heads up, Christian\n\ncarbonated beverage wrote:\n> > Ok thanks for doing that too.\n> > Could you also look at the \"file\" when the test succeed and when it does not \n> > and send us both versions and a diff between them (if it's not too big).\n> \n> Size-wise, they're tiny -- though there's a lot of lines.\n> \n> The diff:\n> \n> ramune/lycaeum:t: diff -u works.t fails.t \n> --- works.t 2007-11-20 21:54:29.000000000 -0700\n> +++ fails.t 2007-11-20 21:54:39.000000000 -0700\n> @@ -55,9 +55,9 @@\n> 55\n> 56\n> 57\n> -5588\n> +58\n> 59\n> 60\n> -6611\n> +61\n> 62\n> 63\n\nAh, these changes should've been made on the svn side. But I'm betting\na race condition appears on faster computers (which we've seen before in\nother tests).\n\nI can't reproduce it at all on any of my boxes, but does the following\none-liner fix it consistently?\n\nThanks,\n\nFrom: Eric Wong \nDate: Wed, 21 Nov 2007 00:57:33 -0800\nSubject: [PATCH] t9106: fix a race condition that caused svn to miss modifications\n\ncarbonated beverage noticed this test was occasionally failing.\n\nSigned-off-by: Eric Wong \n---\n t/t9106-git-svn-dcommit-clobber-series.sh | 1 +\n 1 files changed, 1 insertions(+), 0 deletions(-)\n\ndiff --git a/t/t9106-git-svn-dcommit-clobber-series.sh b/t/t9106-git-svn-dcommit-clobber-series.sh\nindex d59acc8..7452546 100755\n--- a/t/t9106-git-svn-dcommit-clobber-series.sh\n+++ b/t/t9106-git-svn-dcommit-clobber-series.sh\n@@ -22,6 +22,7 @@ test_expect_success '(supposedly) non-conflicting change from SVN' \"\n \tcd tmp &&\n \t\tperl -i -p -e 's/^58\\$/5588/' file &&\n \t\tperl -i -p -e 's/^61\\$/6611/' file &&\n+\t\tpoke file &&\n \t\ttest x\\\"\\`sed -n -e 58p < file\\`\\\" = x5588 &&\n \t\ttest x\\\"\\`sed -n -e 61p < file\\`\\\" = x6611 &&\n \t\tsvn commit -m '58 => 5588, 61 => 6611' &&\n-- \nEric Wong\n"},{"id":"60488","messageId":"20071121095727.GA13211@net-ronin.org","threadId":"10938","inReplyTo":"20071121091001.GA23266@soma","subject":"Re: t9106 failure, bisect weirdness","fromName":"carbonated beverage","fromEmail":"ramune@net-ronin.org","sentAt":"2007-11-21T09:57:27Z","receivedAt":"2007-11-21T09:57:27Z","isPatch":false,"sender":{"key":"ramune@net-ronin.org","avatar":null},"body":"> I can't reproduce it at all on any of my boxes, but does the following\n> one-liner fix it consistently?\n\nFYI, it's a Core 2 Duo @ 2GHz (Thinkpad Z61t).\n\nYup, I ran the test in a loop, and it went through 15 iterations without\nfailure. This patch works for me.\n\nThanks!\n\n-- DN\nDaniel\n"}]}