{"thread":{"id":"25361","subject":"1e633418479926bc85ed21a4f91c845a3dd3ad66 breaks on OSX","startedAt":"2010-10-05T20:48:23Z","lastAt":"2010-10-06T01:45:18Z","messageCount":6,"participants":["Randal L. Schwartz","Ævar Arnfjörð Bjarmason","Jeff King","Sverre Rabbelier"],"isPatch":false,"patchVersion":null,"patchTotal":null},"messages":[{"id":"152721","messageId":"86r5g42wo8.fsf@red.stonehenge.com","threadId":"25361","inReplyTo":null,"subject":"1e633418479926bc85ed21a4f91c845a3dd3ad66 breaks on OSX","fromName":"Randal L. Schwartz","fromEmail":"merlyn@stonehenge.com","sentAt":"2010-10-05T20:48:23Z","receivedAt":"2010-10-05T20:48:23Z","isPatch":false,"sender":{"key":"merlyn@stonehenge.com","avatar":"https://gravatar.com/avatar/dc528d210743ff0333e6213f9ee7b33b23f1b7bc1f3c5a8c2d819074ecd7ab19?d=mp&s=160"},"body":"\n SUBDIR perl\nmake[1]: `perl.mak' is up to date.\n GEN git-add--interactive\nWriting perl.mak for Git\nmake[2]: *** [perl.mak] Error 1\nmake[1]: *** [instlibdir] Error 2\nmake: *** [git-add--interactive] Error 2\n\n-- \nRandal L. Schwartz - Stonehenge Consulting Services, Inc. - +1 503 777 0095\n \nSmalltalk/Perl/Unix consulting, Technical writing, Comedy, etc. etc.\nSee http://methodsandmessages.posterous.com/ for Smalltalk discussion\n"},{"id":"152724","messageId":"AANLkTinq7haFRs6h8W+ycDy5MzFgjty6LaD8syL+KnHX@mail.gmail.com","threadId":"25361","inReplyTo":"86r5g42wo8.fsf@red.stonehenge.com","subject":"Re: 1e633418479926bc85ed21a4f91c845a3dd3ad66 breaks on OSX","fromName":"Ævar Arnfjörð Bjarmason","fromEmail":"avarab@gmail.com","sentAt":"2010-10-05T21:05:45Z","receivedAt":"2010-10-05T21:05:45Z","isPatch":false,"sender":{"key":"avarab@gmail.com","avatar":"https://avatars.githubusercontent.com/u/45301?v=4"},"body":"On Tue, Oct 5, 2010 at 20:48, Randal L. Schwartz wrote:\n>    SUBDIR perl\n> make[1]: `perl.mak' is up to date.\n>    GEN git-add--interactive\n> Writing perl.mak for Git\n> make[2]: *** [perl.mak] Error 1\n> make[1]: *** [instlibdir] Error 2\n> make: *** [git-add--interactive] Error 2\n\n1e63341 is a merge commit, and none of the merged things (in the\ncommit message) seem to have anything to do with the Perl makefile\nprocess.\n\nCan you run:\n\n make -j 1 V=1\n\nAnd report what output you get?\n"},{"id":"152727","messageId":"86iq1g2vhq.fsf@red.stonehenge.com","threadId":"25361","inReplyTo":"AANLkTinq7haFRs6h8W+ycDy5MzFgjty6LaD8syL+KnHX@mail.gmail.com","subject":"Re: 1e633418479926bc85ed21a4f91c845a3dd3ad66 breaks on OSX","fromName":"Randal L. Schwartz","fromEmail":"merlyn@stonehenge.com","sentAt":"2010-10-05T21:13:53Z","receivedAt":"2010-10-05T21:13:53Z","isPatch":false,"sender":{"key":"merlyn@stonehenge.com","avatar":"https://gravatar.com/avatar/dc528d210743ff0333e6213f9ee7b33b23f1b7bc1f3c5a8c2d819074ecd7ab19?d=mp&s=160"},"body":">>>>> \"Ævar\" == Ævar Arnfjörð Bjarmason writes:\n\nÆvar> 1e63341 is a merge commit, and none of the merged things (in the\nÆvar> commit message) seem to have anything to do with the Perl makefile\nÆvar> process.\n\nÆvar> Can you run:\n\nÆvar> make -j 1 V=1\n\nÆvar> And report what output you get?\n\nOddly enough, just rerunning the make again *worked*.\n\nThat's scary.\n\nAnd therefore also not bisectable. :(\n\nI imagine there's some heisenbug in there, that we'll probably trip over\nlater.\n\n-- \nRandal L. Schwartz - Stonehenge Consulting Services, Inc. - +1 503 777 0095\n \nSmalltalk/Perl/Unix consulting, Technical writing, Comedy, etc. etc.\nSee http://methodsandmessages.posterous.com/ for Smalltalk discussion\n"},{"id":"152728","messageId":"20101005213401.GA22845@sigill.intra.peff.net","threadId":"25361","inReplyTo":"86iq1g2vhq.fsf@red.stonehenge.com","subject":"Re: 1e633418479926bc85ed21a4f91c845a3dd3ad66 breaks on OSX","fromName":"Jeff King","fromEmail":"peff@peff.net","sentAt":"2010-10-05T21:34:02Z","receivedAt":"2010-10-05T21:34:02Z","isPatch":false,"sender":{"key":"peff@peff.net","avatar":"https://avatars.githubusercontent.com/u/45925?v=4"},"body":"On Tue, Oct 05, 2010 at 02:13:53PM -0700, Randal L. Schwartz wrote:\n\n> >>>>> \"Ævar\" == Ævar Arnfjörð Bjarmason writes:\n> \n> Ævar> 1e63341 is a merge commit, and none of the merged things (in the\n> Ævar> commit message) seem to have anything to do with the Perl makefile\n> Ævar> process.\n> \n> Ævar> Can you run:\n> \n> Ævar> make -j 1 V=1\n> \n> Ævar> And report what output you get?\n> \n> Oddly enough, just rerunning the make again *worked*.\n> \n> That's scary.\n> \n> And therefore also not bisectable. :(\n> \n> I imagine there's some heisenbug in there, that we'll probably trip over\n> later.\n\nThis has come up before, but nobody has managed to track it down. I did\nsome looking at it in the past, but this is all I could find:\n\n http://article.gmane.org/gmane.comp.version-control.git/117710\n\nIf my analysis there is right, we could probably protect against it with\nsomething like:\n\ndiff --git a/Makefile b/Makefile\nindex 25f94b0..aaef528 100644\n--- a/Makefile\n+++ b/Makefile\n@@ -1596,6 +1596,10 @@ ifndef NO_PYTHON\n endif\n \t$(QUIET_SUBDIR0)templates $(QUIET_SUBDIR1) SHELL_PATH='$(SHELL_PATH_SQ)' PERL_PATH='$(PERL_PATH_SQ)'\n \n+ifndef NO_PERL\n+all:: perl/perl.mak\n+endif\n+\n please_set_SHELL_PATH_to_a_more_modern_shell:\n \t@$$(:)\n \n\nBut of course it's hard to test.\n\n-Peff\n"},{"id":"152733","messageId":"AANLkTimwRJ+X4Kom9_Nv-5Xjf1LKCPbkwavWYHqCdTaM@mail.gmail.com","threadId":"25361","inReplyTo":"20101005213401.GA22845@sigill.intra.peff.net","subject":"Re: 1e633418479926bc85ed21a4f91c845a3dd3ad66 breaks on OSX","fromName":"Sverre Rabbelier","fromEmail":"srabbelier@gmail.com","sentAt":"2010-10-05T22:33:20Z","receivedAt":"2010-10-05T22:33:20Z","isPatch":false,"sender":{"key":"srabbelier@gmail.com","avatar":"https://avatars.githubusercontent.com/u/3098?v=4"},"body":"Heya,\n\nOn Tue, Oct 5, 2010 at 23:34, Jeff King wrote:\n> +ifndef NO_PERL\n> +all:: perl/perl.mak\n> +endif\n> +\n\nRegardless of whether that fixes the problem that looks like a sane\npatch, so maybe we just should apply it?\n\n-- \nCheers,\n\nSverre Rabbelier\n"},{"id":"152739","messageId":"20101006014518.GA8557@sigill.intra.peff.net","threadId":"25361","inReplyTo":"AANLkTimwRJ+X4Kom9_Nv-5Xjf1LKCPbkwavWYHqCdTaM@mail.gmail.com","subject":"Re: 1e633418479926bc85ed21a4f91c845a3dd3ad66 breaks on OSX","fromName":"Jeff King","fromEmail":"peff@peff.net","sentAt":"2010-10-06T01:45:18Z","receivedAt":"2010-10-06T01:45:18Z","isPatch":false,"sender":{"key":"peff@peff.net","avatar":"https://avatars.githubusercontent.com/u/45925?v=4"},"body":"On Wed, Oct 06, 2010 at 12:33:20AM +0200, Sverre Rabbelier wrote:\n\n> On Tue, Oct 5, 2010 at 23:34, Jeff King wrote:\n> > +ifndef NO_PERL\n> > +all:: perl/perl.mak\n> > +endif\n> > +\n> \n> Regardless of whether that fixes the problem that looks like a sane\n> patch, so maybe we just should apply it?\n\nI'm not opposed, though the situation is confusing enough I'd like to\nhave somebody confirm that my previous analysis makes sense (otherwise,\nthis is just adding cruft to the Makefile :) ).\n\nSadly, I can't seem to replicate the original problem, even doing\nsomething like:\n\n n=1\n while true; do\n make clean;\n if ! make -j4; then break; fi\n echo $n; n=$(($n+1))\n done\n\n(and I tried several variations). It ran for 20 minutes without failing.\nMy suspicion is that there may be some particular state that triggers\nthe race condition, but I can't seem to figure it out.\n\n-Peff\n"}]}