{"thread":{"id":"33235","subject":"[PATCH jk/pkt-line-cleanup] t5700-clone-reference: send trace to fd 2, not 3, to please Windows git","startedAt":"2013-03-20T08:24:44Z","lastAt":"2013-03-21T15:04:06Z","messageCount":11,"participants":["Johannes Sixt","Jeff King","Junio C Hamano","Eric Sunshine"],"isPatch":true,"patchVersion":1,"patchTotal":null},"messages":[{"id":"211727","messageId":"5149724C.3060105@viscovery.net","threadId":"33235","inReplyTo":null,"subject":"[PATCH jk/pkt-line-cleanup] t5700-clone-reference: send trace to fd 2, not 3, to please Windows git","fromName":"Johannes Sixt","fromEmail":"j.sixt@viscovery.net","sentAt":"2013-03-20T08:24:44Z","receivedAt":"2013-03-20T08:24:44Z","isPatch":true,"sender":{"key":"j6t@kdbg.org","avatar":"https://avatars.githubusercontent.com/u/14810926?v=4"},"body":"From: Johannes Sixt <j6t@kdbg.org>\n\nTwo tests use GIT_TRACE=3 to dump debugging information of git. On\nWindows, however, bash is unable to set up file descriptor 3 correctly\nfor its child process, so that git reports \"Bad file descriptor\" on\nevery trace attempt. The 'git clone' test succeeds nevertheless because\nan empty trace file remains, and there is only a check for the\nabsence of a particular line. But the 'git fetch' process ultimately\nhangs (the dynamics that lead to this surprising result have not been\ninvestigated).\n\nSince we do not otherwise use stderr in the test cases, divert the\ntrace dump to stderr.\n\nSigned-off-by: Johannes Sixt <j6t@kdbg.org>\n---\n This fixes a regression introduced in \n\n t/t5700-clone-reference.sh | 4 ++--\n 1 file changed, 2 insertions(+), 2 deletions(-)\n\ndiff --git a/t/t5700-clone-reference.sh b/t/t5700-clone-reference.sh\nindex 9cd3b4d..8b5c58e 100755\n--- a/t/t5700-clone-reference.sh\n+++ b/t/t5700-clone-reference.sh\n@@ -55,7 +55,7 @@ cd \"$base_dir\"\n rm -f \"$U.D\"\n \n test_expect_success 'cloning with reference (no -l -s)' \\\n-'GIT_TRACE_PACKET=3 git clone --reference B \"file://$(pwd)/A\" D 3>\"$U.D\"'\n+'GIT_TRACE_PACKET=2 git clone --reference B \"file://$(pwd)/A\" D 2>\"$U.D\"'\n \n test_expect_success 'fetched no objects' \\\n '! grep \" want\" \"$U.D\"'\n@@ -173,7 +173,7 @@ test_expect_success 'fetch with incomplete alternates' '\n \t(\n \t\tcd K &&\n \t\tgit remote add J \"file://$base_dir/J\" &&\n-\t\tGIT_TRACE_PACKET=3 git fetch J 3>\"$U.K\"\n+\t\tGIT_TRACE_PACKET=2 git fetch J 2>\"$U.K\"\n \t) &&\n \tmaster_object=$(cd A && git for-each-ref --format=\"%(objectname)\" refs/heads/master) &&\n \t! grep \" want $master_object\" \"$U.K\" &&\n-- \n1.8.2.1298.g2825a8e\n"},{"id":"211729","messageId":"20130320093313.GA9482@sigill.intra.peff.net","threadId":"33235","inReplyTo":"5149724C.3060105@viscovery.net","subject":"Re: [PATCH jk/pkt-line-cleanup] t5700-clone-reference: send trace to fd 2, not 3, to please Windows git","fromName":"Jeff King","fromEmail":"peff@peff.net","sentAt":"2013-03-20T09:33:13Z","receivedAt":"2013-03-20T09:33:13Z","isPatch":true,"sender":{"key":"peff@peff.net","avatar":"https://avatars.githubusercontent.com/u/45925?v=4"},"body":"On Wed, Mar 20, 2013 at 09:24:44AM +0100, Johannes Sixt wrote:\n\n> From: Johannes Sixt <j6t@kdbg.org>\n> \n> Two tests use GIT_TRACE=3 to dump debugging information of git. On\n> Windows, however, bash is unable to set up file descriptor 3 correctly\n> for its child process, so that git reports \"Bad file descriptor\" on\n> every trace attempt. The 'git clone' test succeeds nevertheless because\n> an empty trace file remains, and there is only a check for the\n> absence of a particular line. But the 'git fetch' process ultimately\n> hangs (the dynamics that lead to this surprising result have not been\n> investigated).\n> \n> Since we do not otherwise use stderr in the test cases, divert the\n> trace dump to stderr.\n\nI think that is OK, but I'm curious why this is a problem _now_, and not\nwith the code prior to 97a83fa8. The old GIT_DEBUG_SEND_PACK was also\njust calling write() to descriptor 3.\n\n> Signed-off-by: Johannes Sixt <j6t@kdbg.org>\n> ---\n>  This fixes a regression introduced in \n\n...in 97a83fa8, I assume.\n\n-Peff\n"},{"id":"211732","messageId":"51499DE7.2030404@viscovery.net","threadId":"33235","inReplyTo":"20130320093313.GA9482@sigill.intra.peff.net","subject":"Re: [PATCH jk/pkt-line-cleanup] t5700-clone-reference: send trace to fd 2, not 3, to please Windows git","fromName":"Johannes Sixt","fromEmail":"j.sixt@viscovery.net","sentAt":"2013-03-20T11:30:47Z","receivedAt":"2013-03-20T11:30:47Z","isPatch":true,"sender":{"key":"j6t@kdbg.org","avatar":"https://avatars.githubusercontent.com/u/14810926?v=4"},"body":"Am 3/20/2013 10:33, schrieb Jeff King:\n> On Wed, Mar 20, 2013 at 09:24:44AM +0100, Johannes Sixt wrote:\n> \n>> From: Johannes Sixt <j6t@kdbg.org>\n>>\n>> Two tests use GIT_TRACE=3 to dump debugging information of git. On\n>> Windows, however, bash is unable to set up file descriptor 3 correctly\n>> for its child process, so that git reports \"Bad file descriptor\" on\n>> every trace attempt. The 'git clone' test succeeds nevertheless because\n>> an empty trace file remains, and there is only a check for the\n>> absence of a particular line. But the 'git fetch' process ultimately\n>> hangs (the dynamics that lead to this surprising result have not been\n>> investigated).\n>>\n>> Since we do not otherwise use stderr in the test cases, divert the\n>> trace dump to stderr.\n> \n> I think that is OK, but I'm curious why this is a problem _now_, and not\n> with the code prior to 97a83fa8. The old GIT_DEBUG_SEND_PACK was also\n> just calling write() to descriptor 3.\n\nBefore this change, both affected commands completed and the trace file\nwas empty. Notice that in both test cases we only check for the absence of\ncertain lines, which is naturally true for an empty file, so that the\ntests pass.\n\nWith the updated code, 'git fetch' hung, which is how I noticed the\nproblem. As I said, I didn't investigate where and why this happens.\n\n>>  This fixes a regression introduced in \n> \n> ...in 97a83fa8, I assume.\n\nYes, correct. (Sorry, I got distracted while I looked up the commit.)\n\n-- Hannes\n"},{"id":"211765","messageId":"20130320170607.GB10752@sigill.intra.peff.net","threadId":"33235","inReplyTo":"51499DE7.2030404@viscovery.net","subject":"Re: [PATCH jk/pkt-line-cleanup] t5700-clone-reference: send trace to fd 2, not 3, to please Windows git","fromName":"Jeff King","fromEmail":"peff@peff.net","sentAt":"2013-03-20T17:06:07Z","receivedAt":"2013-03-20T17:06:07Z","isPatch":true,"sender":{"key":"peff@peff.net","avatar":"https://avatars.githubusercontent.com/u/45925?v=4"},"body":"On Wed, Mar 20, 2013 at 12:30:47PM +0100, Johannes Sixt wrote:\n\n> > I think that is OK, but I'm curious why this is a problem _now_, and not\n> > with the code prior to 97a83fa8. The old GIT_DEBUG_SEND_PACK was also\n> > just calling write() to descriptor 3.\n> \n> Before this change, both affected commands completed and the trace file\n> was empty. Notice that in both test cases we only check for the absence of\n> certain lines, which is naturally true for an empty file, so that the\n> tests pass.\n\nHmm. The code in t5503 is similar, but before my patch used to actually\nuse \"test -s\" to make sure that some trace output was written. Did it\nfail before 97a83fa8 (and does it pass now)?\n\nWe should probably be adjusting t5503, too. And possibly adding back in\nthe \"test -s\" checks (which I removed as redundant, but I guess the\ndouble-checking might have caught a platform error in this case).\n\nAlso, though I think your use of stderr is probably OK, might it be a\nlittle more robust against future output changes to use:\n\n  GIT_TRACE_PACKET=$PWD/$U.D git clone ...\n\nto write directly to the file. The original GIT_DEBUG_SEND_PACK did not\nsupport such niceties, but GIT_TRACE gives them to us for free.\n\n> With the updated code, 'git fetch' hung, which is how I noticed the\n> problem. As I said, I didn't investigate where and why this happens.\n\nThinking on it more, almost certainly what is happening is that fd 3 is\nbeing used for the actual protocol, and we are jamming random bytes over\nit. Since we have changed the bytes, the code is reacting in a different\nway (but the actual reaction doesn't matter; we should stop the cause,\nnot worry about the symptom).\n\n-Peff\n"},{"id":"211770","messageId":"20130320172640.GC10752@sigill.intra.peff.net","threadId":"33235","inReplyTo":"20130320170607.GB10752@sigill.intra.peff.net","subject":"Re: [PATCH jk/pkt-line-cleanup] t5700-clone-reference: send trace to fd 2, not 3, to please Windows git","fromName":"Jeff King","fromEmail":"peff@peff.net","sentAt":"2013-03-20T17:26:40Z","receivedAt":"2013-03-20T17:26:40Z","isPatch":true,"sender":{"key":"peff@peff.net","avatar":"https://avatars.githubusercontent.com/u/45925?v=4"},"body":"On Wed, Mar 20, 2013 at 01:06:07PM -0400, Jeff King wrote:\n\n> On Wed, Mar 20, 2013 at 12:30:47PM +0100, Johannes Sixt wrote:\n> \n> > > I think that is OK, but I'm curious why this is a problem _now_, and not\n> > > with the code prior to 97a83fa8. The old GIT_DEBUG_SEND_PACK was also\n> > > just calling write() to descriptor 3.\n> > \n> > Before this change, both affected commands completed and the trace file\n> > was empty. Notice that in both test cases we only check for the absence of\n> > certain lines, which is naturally true for an empty file, so that the\n> > tests pass.\n> \n> Hmm. The code in t5503 is similar, but before my patch used to actually\n> use \"test -s\" to make sure that some trace output was written. Did it\n> fail before 97a83fa8 (and does it pass now)?\n\nAh, I see. It did fail, and it was marked with NOT_MINGW. I think we can\nfix that now. Patch coming in a moment.\n\n-Peff\n"},{"id":"211771","messageId":"7va9pyf07c.fsf@alter.siamese.dyndns.org","threadId":"33235","inReplyTo":"20130320170607.GB10752@sigill.intra.peff.net","subject":"Re: [PATCH jk/pkt-line-cleanup] t5700-clone-reference: send trace to fd 2, not 3, to please Windows git","fromName":"Junio C Hamano","fromEmail":"gitster@pobox.com","sentAt":"2013-03-20T17:30:31Z","receivedAt":"2013-03-20T17:30:31Z","isPatch":true,"sender":{"key":"gitster@pobox.com","avatar":"https://avatars.githubusercontent.com/u/54884?v=4"},"body":"Jeff King <peff@peff.net> writes:\n\n> Also, though I think your use of stderr is probably OK, might it be a\n> little more robust against future output changes to use:\n>\n>   GIT_TRACE_PACKET=$PWD/$U.D git clone ...\n>\n> to write directly to the file. The original GIT_DEBUG_SEND_PACK did not\n> support such niceties, but GIT_TRACE gives them to us for free.\n\nSounds very sensible.\n"},{"id":"211774","messageId":"20130320174347.GA28548@sigill.intra.peff.net","threadId":"33235","inReplyTo":"20130320172640.GC10752@sigill.intra.peff.net","subject":"[PATCH] do not use GIT_TRACE_PACKET=3 in tests","fromName":"Jeff King","fromEmail":"peff@peff.net","sentAt":"2013-03-20T17:43:47Z","receivedAt":"2013-03-20T17:43:47Z","isPatch":true,"sender":{"key":"peff@peff.net","avatar":"https://avatars.githubusercontent.com/u/45925?v=4"},"body":"Some test scripts use the GIT_TRACE mechanism to dump\ndebugging information to descriptor 3 (and point it to a\nfile using the shell). On Windows, however, bash is unable\nto set up descriptor 3. We do not write our trace to the\nfile, and worse, we may interfere with other operations\nhappening on descriptor 3, causing tests to fail or other\neven behave inconsistently.\n\nPrior to commit 97a83fa (upload-pack: remove packet debugging\nharness), these tests used GIT_DEBUG_SEND_PACK, which only\nsupported output to a descriptor. The tests in t5503 were\nalways broken on Windows, and were marked to be skipped via\nthe NOT_MINGW prerequisite. In t5700, the tests used to pass\nprior to 97a83fa, but only because they were not careful\nenough; because we only grepped the trace file, an empty\nfile looked successful to us. But post-97a83fa, the writing\nto descriptor 3 causes \"git fetch\" to hang (presumably\nbecause we are throwing random bytes into the middle of the\nprotocol).\n\nNow that we are using the GIT_TRACE mechanism, we can\nimprove both scripts by asking git to write directly to a\nfile rather than a descriptor. That fixes the hang in t5700,\nand should allow t5503 to successfully run on Windows.\n\nIn both cases we now also use \"test -s\" to double-check that\nour trace file actually contains output, which should reduce\nthe possibility of an erroneously passing test.\n\nSigned-off-by: Jeff King <peff@peff.net>\n---\nOn top of jk/pkt-line-cleanup.\n\nThis works for me on Linux, but I do not have a mingw system to test\nt5503 on. If my reasoning is right, it should, but one never knows. :)\n\nJohannes, can you report whether it fixes the problem?\n\n t/t5503-tagfollow.sh       | 38 ++++++++++++++++++--------------------\n t/t5700-clone-reference.sh | 14 +++++++++-----\n 2 files changed, 27 insertions(+), 25 deletions(-)\n\ndiff --git a/t/t5503-tagfollow.sh b/t/t5503-tagfollow.sh\nindex d181c96..f30c038 100755\n--- a/t/t5503-tagfollow.sh\n+++ b/t/t5503-tagfollow.sh\n@@ -4,10 +4,6 @@ test_description='test automatic tag following'\n \n . ./test-lib.sh\n \n-if ! test_have_prereq NOT_MINGW; then\n-\tsay \"GIT_TRACE_PACKET not supported - skipping tests\"\n-fi\n-\n # End state of the repository:\n #\n #         T - tag1          S - tag2\n@@ -17,7 +13,7 @@ fi\n #     \\   C - origin/cat    \\\n #      origin/master         master\n \n-test_expect_success NOT_MINGW setup '\n+test_expect_success setup '\n \ttest_tick &&\n \techo ichi >file &&\n \tgit add file &&\n@@ -39,33 +35,35 @@ test_expect_success NOT_MINGW 'fetch A (new commit : 1 connection)' '\n '\n \n U=UPLOAD_LOG\n+UPATH=\"$(pwd)/$U\"\n \n-test_expect_success NOT_MINGW 'setup expect' '\n+test_expect_success 'setup expect' '\n cat - <<EOF >expect\n want $A\n EOF\n '\n \n get_needs () {\n+\ttest -s \"$1\" &&\n \tperl -alne '\n \t\tnext unless $F[1] eq \"upload-pack<\";\n \t\tlast if $F[2] eq \"0000\";\n \t\tprint $F[2], \" \", $F[3];\n-\t' \"$@\"\n+\t' \"$1\"\n }\n \n-test_expect_success NOT_MINGW 'fetch A (new commit : 1 connection)' '\n+test_expect_success 'fetch A (new commit : 1 connection)' '\n \trm -f $U &&\n \t(\n \t\tcd cloned &&\n-\t\tGIT_TRACE_PACKET=3 git fetch 3>../$U &&\n+\t\tGIT_TRACE_PACKET=$UPATH git fetch &&\n \t\ttest $A = $(git rev-parse --verify origin/master)\n \t) &&\n \tget_needs $U >actual &&\n \ttest_cmp expect actual\n '\n \n-test_expect_success NOT_MINGW \"create tag T on A, create C on branch cat\" '\n+test_expect_success \"create tag T on A, create C on branch cat\" '\n \tgit tag -a -m tag1 tag1 $A &&\n \tT=$(git rev-parse --verify tag1) &&\n \n@@ -77,18 +75,18 @@ test_expect_success NOT_MINGW 'fetch C, T (new branch, tag : 1 connection)' '\n \tgit checkout master\n '\n \n-test_expect_success NOT_MINGW 'setup expect' '\n+test_expect_success 'setup expect' '\n cat - <<EOF >expect\n want $C\n want $T\n EOF\n '\n \n-test_expect_success NOT_MINGW 'fetch C, T (new branch, tag : 1 connection)' '\n+test_expect_success 'fetch C, T (new branch, tag : 1 connection)' '\n \trm -f $U &&\n \t(\n \t\tcd cloned &&\n-\t\tGIT_TRACE_PACKET=3 git fetch 3>../$U &&\n+\t\tGIT_TRACE_PACKET=$UPATH git fetch &&\n \t\ttest $C = $(git rev-parse --verify origin/cat) &&\n \t\ttest $T = $(git rev-parse --verify tag1) &&\n \t\ttest $A = $(git rev-parse --verify tag1^0)\n@@ -97,7 +95,7 @@ test_expect_success NOT_MINGW 'fetch C, T (new branch, tag : 1 connection)' '\n \ttest_cmp expect actual\n '\n \n-test_expect_success NOT_MINGW \"create commits O, B, tag S on B\" '\n+test_expect_success \"create commits O, B, tag S on B\" '\n \ttest_tick &&\n \techo O >file &&\n \tgit add file &&\n@@ -113,18 +111,18 @@ test_expect_success NOT_MINGW 'fetch B, S (commit and tag : 1 connection)' '\n \tS=$(git rev-parse --verify tag2)\n '\n \n-test_expect_success NOT_MINGW 'setup expect' '\n+test_expect_success 'setup expect' '\n cat - <<EOF >expect\n want $B\n want $S\n EOF\n '\n \n-test_expect_success NOT_MINGW 'fetch B, S (commit and tag : 1 connection)' '\n+test_expect_success 'fetch B, S (commit and tag : 1 connection)' '\n \trm -f $U &&\n \t(\n \t\tcd cloned &&\n-\t\tGIT_TRACE_PACKET=3 git fetch 3>../$U &&\n+\t\tGIT_TRACE_PACKET=$UPATH git fetch &&\n \t\ttest $B = $(git rev-parse --verify origin/master) &&\n \t\ttest $B = $(git rev-parse --verify tag2^0) &&\n \t\ttest $S = $(git rev-parse --verify tag2)\n@@ -133,14 +131,14 @@ EOF\n \ttest_cmp expect actual\n '\n \n-test_expect_success NOT_MINGW 'setup expect' '\n+test_expect_success 'setup expect' '\n cat - <<EOF >expect\n want $B\n want $S\n EOF\n '\n \n-test_expect_success NOT_MINGW 'new clone fetch master and tags' '\n+test_expect_success 'new clone fetch master and tags' '\n \tgit branch -D cat\n \trm -f $U\n \t(\n@@ -148,7 +146,7 @@ test_expect_success NOT_MINGW 'new clone fetch master and tags' '\n \t\tcd clone2 &&\n \t\tgit init &&\n \t\tgit remote add origin .. &&\n-\t\tGIT_TRACE_PACKET=3 git fetch 3>../$U &&\n+\t\tGIT_TRACE_PACKET=$UPATH git fetch &&\n \t\ttest $B = $(git rev-parse --verify origin/master) &&\n \t\ttest $S = $(git rev-parse --verify tag2) &&\n \t\ttest $B = $(git rev-parse --verify tag2^0) &&\ndiff --git a/t/t5700-clone-reference.sh b/t/t5700-clone-reference.sh\nindex 9cd3b4d..60f1552 100755\n--- a/t/t5700-clone-reference.sh\n+++ b/t/t5700-clone-reference.sh\n@@ -54,11 +54,14 @@ test_expect_success 'cloning with reference (no -l -s)' \\\n \n rm -f \"$U.D\"\n \n-test_expect_success 'cloning with reference (no -l -s)' \\\n-'GIT_TRACE_PACKET=3 git clone --reference B \"file://$(pwd)/A\" D 3>\"$U.D\"'\n+test_expect_success 'cloning with reference (no -l -s)' '\n+\tGIT_TRACE_PACKET=$U.D git clone --reference B \"file://$(pwd)/A\" D\n+'\n \n-test_expect_success 'fetched no objects' \\\n-'! grep \" want\" \"$U.D\"'\n+test_expect_success 'fetched no objects' '\n+\ttest -s \"$U.D\" &&\n+\t! grep \" want\" \"$U.D\"\n+'\n \n cd \"$base_dir\"\n \n@@ -173,9 +176,10 @@ test_expect_success 'fetch with incomplete alternates' '\n \t(\n \t\tcd K &&\n \t\tgit remote add J \"file://$base_dir/J\" &&\n-\t\tGIT_TRACE_PACKET=3 git fetch J 3>\"$U.K\"\n+\t\tGIT_TRACE_PACKET=$U.K git fetch J\n \t) &&\n \tmaster_object=$(cd A && git for-each-ref --format=\"%(objectname)\" refs/heads/master) &&\n+\ttest -s \"$U.K\" &&\n \t! grep \" want $master_object\" \"$U.K\" &&\n \ttag_object=$(cd A && git for-each-ref --format=\"%(objectname)\" refs/tags/HEAD) &&\n \t! grep \" want $tag_object\" \"$U.K\"\n-- \n1.8.2.22.g1efe1a3\n"},{"id":"211822","messageId":"CAPig+cQhG-oE5n-RjxhHqGoPzTZbLGagU+9gxzrJhaOcUAtffQ@mail.gmail.com","threadId":"33235","inReplyTo":"20130320174347.GA28548@sigill.intra.peff.net","subject":"Re: [PATCH] do not use GIT_TRACE_PACKET=3 in tests","fromName":"Eric Sunshine","fromEmail":"sunshine@sunshineco.com","sentAt":"2013-03-20T20:29:40Z","receivedAt":"2013-03-20T20:29:40Z","isPatch":true,"sender":{"key":"sunshine@sunshineco.com","avatar":"https://avatars.githubusercontent.com/u/163641?v=4"},"body":"On Wed, Mar 20, 2013 at 1:43 PM, Jeff King <peff@peff.net> wrote:\n> Some test scripts use the GIT_TRACE mechanism to dump\n> debugging information to descriptor 3 (and point it to a\n> file using the shell). On Windows, however, bash is unable\n> to set up descriptor 3. We do not write our trace to the\n> file, and worse, we may interfere with other operations\n> happening on descriptor 3, causing tests to fail or other\n> even behave inconsistently.\n\ns/other even/even/\n"},{"id":"211828","messageId":"20130320205836.GA1373@sigill.intra.peff.net","threadId":"33235","inReplyTo":"CAPig+cQhG-oE5n-RjxhHqGoPzTZbLGagU+9gxzrJhaOcUAtffQ@mail.gmail.com","subject":"Re: [PATCH] do not use GIT_TRACE_PACKET=3 in tests","fromName":"Jeff King","fromEmail":"peff@peff.net","sentAt":"2013-03-20T20:58:36Z","receivedAt":"2013-03-20T20:58:36Z","isPatch":true,"sender":{"key":"peff@peff.net","avatar":"https://avatars.githubusercontent.com/u/45925?v=4"},"body":"On Wed, Mar 20, 2013 at 04:29:40PM -0400, Eric Sunshine wrote:\n\n> On Wed, Mar 20, 2013 at 1:43 PM, Jeff King <peff@peff.net> wrote:\n> > Some test scripts use the GIT_TRACE mechanism to dump\n> > debugging information to descriptor 3 (and point it to a\n> > file using the shell). On Windows, however, bash is unable\n> > to set up descriptor 3. We do not write our trace to the\n> > file, and worse, we may interfere with other operations\n> > happening on descriptor 3, causing tests to fail or other\n> > even behave inconsistently.\n> \n> s/other even/even/\n\nUrgh, I should probably add another proofreading pass to my commit\nmessages. Thanks.\n\n-Peff\n"},{"id":"211839","messageId":"514AAA5D.4000803@viscovery.net","threadId":"33235","inReplyTo":"20130320174347.GA28548@sigill.intra.peff.net","subject":"Re: [PATCH] do not use GIT_TRACE_PACKET=3 in tests","fromName":"Johannes Sixt","fromEmail":"j.sixt@viscovery.net","sentAt":"2013-03-21T06:36:13Z","receivedAt":"2013-03-21T06:36:13Z","isPatch":true,"sender":{"key":"j6t@kdbg.org","avatar":"https://avatars.githubusercontent.com/u/14810926?v=4"},"body":"Am 3/20/2013 18:43, schrieb Jeff King:\n> Now that we are using the GIT_TRACE mechanism, we can\n> improve both scripts by asking git to write directly to a\n> file rather than a descriptor. That fixes the hang in t5700,\n> and should allow t5503 to successfully run on Windows.\n\nWell spotted, and, right, both tests pass with this patch.\n\nTested-by: Johannes Sixt <j6t@kdbg.org>\n\nThanks,\n-- Hannes\n"},{"id":"211871","messageId":"7vli9gbxqx.fsf@alter.siamese.dyndns.org","threadId":"33235","inReplyTo":"514AAA5D.4000803@viscovery.net","subject":"Re: [PATCH] do not use GIT_TRACE_PACKET=3 in tests","fromName":"Junio C Hamano","fromEmail":"gitster@pobox.com","sentAt":"2013-03-21T15:04:06Z","receivedAt":"2013-03-21T15:04:06Z","isPatch":true,"sender":{"key":"gitster@pobox.com","avatar":"https://avatars.githubusercontent.com/u/54884?v=4"},"body":"Johannes Sixt <j.sixt@viscovery.net> writes:\n\n> Am 3/20/2013 18:43, schrieb Jeff King:\n>> Now that we are using the GIT_TRACE mechanism, we can\n>> improve both scripts by asking git to write directly to a\n>> file rather than a descriptor. That fixes the hang in t5700,\n>> and should allow t5503 to successfully run on Windows.\n>\n> Well spotted, and, right, both tests pass with this patch.\n>\n> Tested-by: Johannes Sixt <j6t@kdbg.org>\n\nThanks.\n"}]}