149107 2015-09-13 12:45:50 -0700 IndexedDB does not throw ConstraintErrors for unique keys 2016-03-21 13:57:41 -0700 1 1 1 Unclassified WebKit JavaScriptCore WebKit Nightly Build Mac OS X 10.10 RESOLVED FIXED P2 Normal --- 149117 1 nolan webkit-unassigned beidson info oldest_to_newest 1125493 0 261088 nolan 2015-09-13 12:45:50 -0700 Created attachment 261088 Reproducible test case (May be a dup of https://bugs.webkit.org/show_bug.cgi?id=60720, but it was marked as invalid because they didn't have a good reproducible test case.) In WebKit Nightly 600.8.9 (as well as Safari 8.0.8), it is possible to put() two documents with the same key, even if that key is defined as a unique index. I have a reproducible test case to demonstrate: http://bl.ocks.org/nolanlawson/28ff2db3f3cb386c8ecb. The same test is also attached to this bug. If the bug has been successfully reproduced, then this HTML page will print "There is a bug." Note that openKeyCursor() also returns the same key twice. This bug cannot be reproduced in either Firefox or Chrome. 1171044 1 info 2016-03-04 14:24:57 -0800 +1 for this bug. Reproducable. 1171053 2 beidson 2016-03-04 14:42:52 -0800 Good thing this is fixed in Modern IDB! Works great in ToT/WebKit nightly. 1171055 3 info 2016-03-04 14:43:58 -0800 Will Safari9 in OSX 10.11 get this update? 1171058 4 beidson 2016-03-04 14:46:45 -0800 (In reply to comment #3) > Will Safari9 in OSX 10.11 get this update? This is the WebKit open source project. Safari is an Apple product, and Apple generally does not comment on their future products. 1171062 5 info 2016-03-04 14:50:53 -0800 Generally !== always, huh? :) Not in this Safari version, right? 1171072 6 beidson 2016-03-04 14:58:20 -0800 (In reply to comment #5) > Generally !== always, huh? :) > Not in this Safari version, right? I'm not sure what you're trying to clarify. This should be working for you in whatever relatively modern Safari version you have + a WebKit nightly. 1171082 7 info 2016-03-04 15:29:48 -0800 Okay, thank you 1176743 8 nolan 2016-03-21 08:29:57 -0700 Nice to see progress getting made on this! :) Unfortunately I tested my demo site (http://bl.ocks.org/nolanlawson/28ff2db3f3cb386c8ecb) on WebKit Nightly (10601.5.17.1.1, r195011) and it seems to still be broken. (It says "there is a bug.") Was this closed too early or is there some other reason the test case is failing? 1176753 9 beidson 2016-03-21 09:30:35 -0700 (In reply to comment #8) > Was this closed too early or is there some other reason the test > case is failing? Unclear. I see the same with the attached test case (but did not on 3/4) I'll poke around using the attached test case. 1176790 10 beidson 2016-03-21 11:21:43 -0700 Odd. I reproduced this in the latest WebKit nightly, but it works fine in my local build of ToT. 1176794 11 beidson 2016-03-21 11:27:02 -0700 (In reply to comment #10) > Odd. > > I reproduced this in the latest WebKit nightly, but it works fine in my > local build of ToT. Nope, bonehead - I did *not* reproduce with a WebKit nightly. I was using stock Safari. So, to clarify - I cannot reproduce this being broken. In nightly 198484 (http://builds.nightly.webkit.org/files/trunk/mac/WebKit-SVN-r198484.dmg) and a ToT debug build, the attached test case works fine. ---- If you see the text There is a bug, then there's a bug. Got a ConstraintError when we tried to put() the same key twice Found keys: a Found primaryKeys: 1 1176795 12 beidson 2016-03-21 11:30:37 -0700 (In reply to comment #8) Oh I see: > Unfortunately I tested my demo site > (http://bl.ocks.org/nolanlawson/28ff2db3f3cb386c8ecb) on WebKit Nightly > (10601.5.17.1.1, r195011) and it seems to still be broken. (It says "there > is a bug.") Was this closed too early or is there some other reason the test > case is failing? Nightly 195011 is ancient. January 14th was well before the new engine was enabled in WebKit2, which is what Safari (and therefore WebKit nightlies) uses. Try with the latest nightly, please. 1176796 13 beidson 2016-03-21 11:32:00 -0700 (In reply to comment #12) > (In reply to comment #8) > > Oh I see: > > > Unfortunately I tested my demo site > > (http://bl.ocks.org/nolanlawson/28ff2db3f3cb386c8ecb) on WebKit Nightly > > (10601.5.17.1.1, r195011) and it seems to still be broken. (It says "there > > is a bug.") Was this closed too early or is there some other reason the test > > case is failing? > > Nightly 195011 is ancient. > > January 14th was well before the new engine was enabled in WebKit2, which is > what Safari (and therefore WebKit nightlies) uses. > > Try with the latest nightly, please. For the historical record: Modern IDB wasn't enabled in WK2 until http://trac.webkit.org/changeset/197122, and there were still some pretty bad bugs with it for another ~1 week after that. 1176841 14 nolan 2016-03-21 13:57:41 -0700 My bad! It is indeed fixed. Somehow I got stuck with an old version of WebKit Nightly, but re-downloading worked. :) I'm on r198490. Also a big thank-you for your work on "modern IDB"! Love seeing all these IDB bugs slowly getting knocked down. 261088 2015-09-13 12:45:50 -0700 2015-09-13 12:45:50 -0700 Reproducible test case index.html text/html 1901 nolan PGJvZHk+CiAgPGgxPiBSZXBybyBXZWJLaXQgSW5kZXhlZERCIHVuaXF1ZSBpbmRleCBidWcgPC9o MT4KICA8cD5JZiB5b3Ugc2VlIHRoZSB0ZXh0IDxzdHJvbmc+VGhlcmUgaXMgYSBidWc8L3N0cm9u Zz4sIHRoZW4gdGhlcmUncyBhIGJ1Zy48L3A+CjxkaXYgaWQ9ImxvZ2dlciI+PC9kaXY+CjxzY3Jp cHQ+Cgp2YXIgZGJuYW1lID0gJ3JhbmRvXycgKyBNYXRoLnJhbmRvbSgpICsgJ18nOwoKZnVuY3Rp b24gbG9nKG1zZykgewogIGRvY3VtZW50LmdldEVsZW1lbnRCeUlkKCJsb2dnZXIiKS5pbm5lckhU TUwgKz0gbXNnICsgIjxicj4iOwp9Cgp2YXIgcmVxID0gaW5kZXhlZERCLm9wZW4oZGJuYW1lICsg IjEiKTsKCnJlcS5vbnVwZ3JhZGVuZWVkZWQgPSBmdW5jdGlvbiAoZSkgewogIHZhciBkYiA9IGUu dGFyZ2V0LnJlc3VsdDsKCiAgaWYgKGUub2xkVmVyc2lvbiA9PT0gMSkgewogICAgcmV0dXJuOwog IH0KCiAgZGIuY3JlYXRlT2JqZWN0U3RvcmUoJ3N0b3JlJywge2F1dG9JbmNyZW1lbnQ6IHRydWV9 KS5jcmVhdGVJbmRleCgna2V5JywgJ2tleScsIHt1bmlxdWU6IHRydWV9KTsKfTsKCnJlcS5vbnN1 Y2Nlc3MgPSBmdW5jdGlvbiAoZSkgewogIHZhciBkYiA9IGUudGFyZ2V0LnJlc3VsdDsKICAKICB2 YXIgdHhuID0gZGIudHJhbnNhY3Rpb24oJ3N0b3JlJywgJ3JlYWR3cml0ZScpOwogIAogIHR4bi5v YmplY3RTdG9yZSgnc3RvcmUnKS5wdXQoe2tleTogJ2EnfSkKICAKICB0eG4ub25jb21wbGV0ZSA9 IGZ1bmN0aW9uICgpIHsKICAgIHZhciB0eG4gPSBkYi50cmFuc2FjdGlvbignc3RvcmUnLCAncmVh ZHdyaXRlJyk7CiAgICAKICAgIHZhciBwdXRSZXEgPSB0eG4ub2JqZWN0U3RvcmUoJ3N0b3JlJyku cHV0KHtrZXk6ICdhJ30pOwogICAgcHV0UmVxLm9uc3VjY2VzcyA9IGZ1bmN0aW9uICgpIHsKICAg ICAgbG9nKCdXZSBtYW5hZ2VkIHRvIHB1dCgpIHRoZSBzYW1lIGtleSB0d2ljZSEnKQogICAgICBs b2coJzxzdHJvbmc+VGhlcmUgaXMgYSBidWc8L3N0cm9uZz4nKQogICAgfTsKICAgIHB1dFJlcS5v bmVycm9yID0gZnVuY3Rpb24gKGUpIHsKICAgICAgbG9nKCdHb3QgYSBDb25zdHJhaW50RXJyb3Ig d2hlbiB3ZSB0cmllZCB0byBwdXQoKSB0aGUgc2FtZSBrZXkgdHdpY2UnKTsKICAgICAgZS5wcmV2 ZW50RGVmYXVsdCgpOyAvLyBhdm9pZCB0cmFuc2FjdGlvbiBhYm9ydAogICAgICBlLnN0b3BQcm9w YWdhdGlvbigpOyAvLyBhdm9pZCB0cmFuc2FjdGlvbiBvbmVycm9yCiAgICB9OwogICAgICAgIAog ICAgdHhuLm9uY29tcGxldGUgPSBmdW5jdGlvbiAoKSB7CiAgICAKICAgICAgdmFyIHR4biA9IGRi LnRyYW5zYWN0aW9uKCdzdG9yZScsICdyZWFkb25seScpOwogICAgICB2YXIga2V5cyA9IFtdOwog ICAgICB2YXIgcHJpbWFyeUtleXMgPSBbXTsKICAgICAgdHhuLm9iamVjdFN0b3JlKCdzdG9yZScp LmluZGV4KCdrZXknKS5vcGVuS2V5Q3Vyc29yKCkub25zdWNjZXNzID0gZnVuY3Rpb24gKGUpIHsK ICAgICAgICB2YXIgY3Vyc29yID0gZS50YXJnZXQucmVzdWx0OwogICAgICAgIGlmICghY3Vyc29y KSB7CiAgICAgICAgICByZXR1cm47CiAgICAgICAgfQogIAogICAgICAgIGtleXMucHVzaChjdXJz b3Iua2V5KTsKICAgICAgICBwcmltYXJ5S2V5cy5wdXNoKGN1cnNvci5wcmltYXJ5S2V5KTsKICAK ICAgICAgICBjdXJzb3IuY29udGludWUoKTsKICAgICAgfTsKICAgICAgdHhuLm9uY29tcGxldGUg PSBmdW5jdGlvbiAoKSB7CiAgICAgICAgbG9nKCcnKQogICAgICAgIGxvZygnRm91bmQga2V5czog JyArIGtleXMuam9pbignLCAnKSk7CiAgICAgICAgbG9nKCdGb3VuZCBwcmltYXJ5S2V5czogJyAr IHByaW1hcnlLZXlzLmpvaW4oJywgJykpOwogICAgICB9OwogICAgfTsKICB9Owp9OwoKPC9zY3Jp cHQ+CjwvYm9keT4KPC9odG1sPgo=