3539 2005-06-14 19:19:33 -0700 Array join and toString methods do not support circular references 2005-12-10 11:04:16 -0800 1 1 1 Unclassified WebKit JavaScriptCore 412 Mac OS X 10.4 RESOLVED FIXED P2 Normal --- 1 david oliver brendan oliver oldest_to_newest 12055 0 david 2005-06-14 19:19:33 -0700 If I have a circular reference in WebKit JavaScript, I can't dig into it effectively. For example, given this code: var ary = [1, 2]; ary.push(ary); for (var i = 0; i < ary.length; i++) { document.write(typeof ary[0] + ': ' + ary[i] + ' '); } In Safari, this outputs "number: 1 number: 2". In Firefox, it outputs "number: 1 number: 2 number: 1,2, ". The upshot is that I cannot access any of the elements of ary when I get at it through ary[2], but I should be able to. ary == ary[2] does evaluate to true, so I'm not sure what's happening here. 12067 1 2348 joost 2005-06-14 23:27:28 -0700 Created attachment 2348 testcase 12068 2 joost 2005-06-14 23:28:21 -0700 I'll confirm this one, since it doesn't do the same as it does in Firefox either. But, please attach the testcase yourself next time :). 12478 3 david 2005-06-17 11:54:35 -0700 I forgot to mention that the reason it does not print what's in the third slot of the array is that an exception is thrown. The exception is: TypeError – No default value undefined Line: 0 My guess is that this is triggered somehow by the circular reference. Also, this only comes up if I iterate over the array using for() or do something else like call sort(). If I just reach into the circular reference, like this: document.write(ary[2][1]); things work fine. HTH. 13130 4 oliver 2005-06-25 06:29:56 -0700 Righto have been going through ECMA 262 document, and it looks like mozilla/firefox's result is non standard. That said the current error given is also not ideal, and that's a result of ObjectImp::defaultValue() unconditionally blatting an exception. I'm attaching a patch that fixes the behaviour of defaultValue() so that the user gets a RangeError (specifically for a stack overflow, the root cause of this problem). The bug is made apparent in the Array.join method which is defined to produce a string containing a list of each element of the array. If an element is undefined or null then an empty string is substituted. Logically in the event of an exception the exception should be propogated (the spec doesn't say otherwise), so that's what we do. Is this an ideal solution? It's what the standard says, but is it what is expected? 13131 5 2640 oliver 2005-06-25 06:32:11 -0700 Created attachment 2640 Patch to fix the *type* of error given to what it should be Fixes the behaviour of ObjectImp::defaultValue() to prevent it from overriding an already thrown exception 13142 6 david 2005-06-25 10:25:28 -0700 Can you cite the passage(s) in the ECSM 262 standard that demonstrate that Firefox/Mozilla's behavior is non-standard? It seems to me that any first-class programming language should properly support circular references. 13143 7 brendan 2005-06-25 11:32:28 -0700 ECMA-262 does not specify behavior in the case of divergence through ToString. This is a flaw in the standard. Edition 3 section 16 allows extensions such as the one in Mozilla's SpiderMonkey engine. In 1998 or so, leading up to Edition 3 work, I implemented and proposed toSource and sharp variables (after Common Lisp) to handle cycles and join points in object graphs. Example from this bug: $ Linux_All_DBG.OBJ/js js> a = [1, 2] 1,2 js> a.push(a) 3 js> a 1,2, js> a.toSource() #1=[1, 2, #1#] In general, in SpiderMonkey, uneval(v) produces a string s such that eval(s) is either === v if v is primitive, or an isomorphic object graph if v is an object reference. In the object case, uneval tries to call toSource, just as toString is tried from [[DefaultValue]] in 262. Some on the committee were against any such complete serialization notation, claiming serialization was too domain-specific to capture in the standard. That is a cop-out: standards follow the 80/20 rule too, or should. Do cycles and join points fall into the hard-case 20 percent? I think not. Does serializing hooks for private native data fall there? Probably, but that's a different case from cycles and join points. There may also have been aesthetic objections, without counterproposals. This is all water under the bridge, unless some subset of ECMA TG1 rallies again for Edition 4. Hope this helps, /be 13165 8 oliver 2005-06-25 22:10:41 -0700 The join operator (what an arrays ToString is implemented as) is definied in ECMA-262 15.4.4.5 for easy reference: <quote> Ar r a y . pr o t o t y pe . j o i n ( s e pa r a t o r ) The el ement s of t he ar r ay ar e conver t ed t o st r i ngs, and t hese st r i ngs ar e t hen concat enat ed, separ at ed by occur r ences of t he separat or. I f no separ at or i s pr ovi ded, a si ngl e comma i s used as t he separ at or . The join met hod t akes one ar gument , separat or, and per f or ms t he f ol l owi ng st eps: 1. Cal l t he [ [ Get ]] met hod of t hi s obj ect wi t h ar gument "length". 2. Cal l ToUi nt 32( Resul t ( 1) ) . 3. I f separat or i s undef i ned, l et separat or be t he si ngl e- char act er st r i ng ",". 4. Cal l ToSt r i ng( separat or) . 5. If Resul t (2) i s zero, ret urn t he empt y st ri ng. 6. Cal l t he [ [ Get ]] met hod of t hi s obj ect wi t h ar gument "0". 7. If Resul t (6) i s undef i ned or nul l , use t he empt y st r i ng; ot her wi se, cal l ToSt r i ng( Resul t ( 6) ) . 8. Let R be Resul t ( 7) . 9. Let k be 1. 10. I f k equal s Resul t ( 2) , r et ur n R. 11. Let S be a st r i ng val ue pr oduced by concat enat i ng R and Resul t ( 4) . 12. Cal l t he [ [ Get ]] met hod of t hi s obj ect wi t h ar gument ToSt r i ng( k) . 13. I f Resul t ( 12) i s undef i ned or nul l , use t he empt y st r i ng; ot her wi se, cal l ToSt r i ng( Resul t ( 12) ) . 14. Let R be a st r i ng val ue pr oduced by concat enat i ng S and Resul t ( 13) . 15. I ncr ease k by 1. 16. Go t o st ep 10. The length pr oper t y of t he join met hod i s 1. NOTE The join f unct i on i s i nt ent i onal l y generi c; i t does not requi re t hat i t s t hi s val ue be an Array obj ect . Theref ore, i t can be t ransf erred t o ot her ki nds of obj ect s f or use as a met hod. Whet her t he join f unct i on can be appl i ed successf ul l y t o a host obj ect i s i mpl ement at i on- dependent . </quote> Given it does not specifically handle exceptions then tthe behaviour should match standard programming semantics -- the exception should be propogated. Brendan: does mozilla just track each object ToString/join has been called in? or does it have a more elegant way of killing off self-recursion? David: JavaScript handles the self recursion easily -- ToString is a recursive function, and it behaves in exactly the same way any language would behave in this case... 13166 9 oliver 2005-06-25 22:21:11 -0700 Given firefox/ie behave the same, and conversation in #webkit, will provide a fix to match firefox/IE behaviour 13197 10 2649 oliver 2005-06-25 23:21:38 -0700 Created attachment 2649 Patch to match behaviour of firefox and IE Okay this patch is a somewhat inelegant solution, but it matches the behaviour of firefox and IE I'm not obsoleting the old patch as a i believe it does fix a bug anyway 13201 11 2651 oliver 2005-06-26 00:19:43 -0700 Created attachment 2651 edge case test for patch sample file demonstrating an array of depth larger than the stack allows -- people running firefox/mac may want to be careful on this one... firefox doesn't seem to like it at all. In this case we just get a straight stack overflow which raises the question of what we should do if toString encounters an exception... 13202 12 brendan 2005-06-26 01:03:41 -0700 (In reply to comment #8) > 13. If Result (12) is undefined or null, use the empty string; otherwise, > call ToString(Result(12)). [snip] > Given it does not specifically handle exceptions then the behaviour should > match standard programming semantics -- the exception should be propogated. Section 16, the second item in the second unordered list, allows extensions to conceal errors. It's also striking that runaway recursion under toString is not dealt with by the spec -- that's a flaw, again, IMO. > Brendan: does mozilla just track each object ToString/join has been called in? > or does it have a more elegant way of killing off self-recursion? The only way to implement this correctly is to map all objects reached by the depth-first search being done by join or toString, using storage local to the current activation to hold the map. See http://lxr.mozilla.org/mozilla/search?string=EnterSharpObject. /be 13203 13 2652 oliver 2005-06-26 01:12:40 -0700 Created attachment 2652 Another edge case, slightly less ugly than previous This is another test case, partly to see what firefox does if an exception occurs during Array.toString firefox/mac doesn't behave well with this one either 13204 14 brendan 2005-06-26 01:57:06 -0700 Firefox 1.0.3 on Linux behaves well enough on those tests, considering. Deer Park alpha 1 on Linux fails immediately with out of memory from document.write -- I'm not sure why. /be 13294 15 2665 oliver 2005-06-27 04:54:20 -0700 Created attachment 2665 test case for cyclic references in array.toString test case that works within the webkit test suite 13295 16 2666 oliver 2005-06-27 04:56:11 -0700 Created attachment 2666 expected results of cyclic reference testcase expected results for webkit testcase 13302 17 2669 oliver 2005-06-27 05:30:22 -0700 Created attachment 2669 testcase for tostring with non-cyclic, but deep hierarchies This test requires the first patch (which allows the correct exception to be propogated) 13303 18 2670 oliver 2005-06-27 05:31:38 -0700 Created attachment 2670 expected output from the deep hierarchies test 13304 19 oliver 2005-06-27 05:35:26 -0700 Should the two patches be merged into one? Strictly speaking the first patch fixes the type of exception the user would see, so might not be considered part of this bug... on the otherhand it does obscure the actual error that has occured. Otherwise if no one has any complaints i'll flag these for review 13326 20 david 2005-06-27 10:19:25 -0700 FWIW, I saw this bug with .sort(), too. If that just uses toString(), then it's the same bug. 13373 21 oliver 2005-06-27 15:35:03 -0700 I did a quick test, and sort seems to work fine -- you sure it wasn't a cycle in the array causing toString to die? With the patches applied it doesn't crash and burn at all, which seem to imply that any problems were in the output... 13375 22 2676 david 2005-06-27 15:45:45 -0700 Created attachment 2676 sort() Test Case No, I'm not sure, but this new test case demonstrates the problem using sort(). If it's just failing because sort uses toString() internally, then great. If not, then is it related at all? 13376 23 oliver 2005-06-27 15:57:25 -0700 Yep that passes with the patches applied -- so sort is using either array.tostring or array.join, either way it works. 13386 24 2640 darin 2005-06-27 17:24:59 -0700 Comment on attachment 2640 Patch to fix the *type* of error given to what it should be Looks like a good change, but it's not formatted correctly. Please look at <http://webkit.opendarwin.org/coding/coding-style.html>. Also, it would be better if this patch had a separate bug report, and a test case demonstrating it's fixed, rather than being part of the "does not support circular references" bug. 13387 25 2649 darin 2005-06-27 17:36:28 -0700 Comment on attachment 2649 Patch to match behaviour of firefox and IE Looks like this is on the right track. Formatting is not really right <http://webkit.opendarwin.org/coding/coding-style.html>. Before landing this, we need to figure out exactly which tests to land with it; at the moment there are lots of different tests attached to the bug and they need to be put in layout test form. I'm not comfortable with using a giant global array as part of the solution here. I suggest we make a small fixed-size global array, then switch over to using a hash table instead once we pass a certain recursion threshold to avoid O(n^2) behavior. That also has the pleasant quality of making the global much smaller. Is it OK to share a single array between both Join and ToString? I'd like to see a test that checks for that. 13395 26 2670 oliver 2005-06-27 18:41:33 -0700 Comment on attachment 2670 expected output from the deep hierarchies test Moved to bug 3743 13396 27 2669 oliver 2005-06-27 18:42:38 -0700 Comment on attachment 2669 testcase for tostring with non-cyclic, but deep hierarchies Moved to bug 3743 13397 28 2640 oliver 2005-06-27 18:44:40 -0700 Comment on attachment 2640 Patch to fix the *type* of error given to what it should be Moved to bug 3743 13398 29 david 2005-06-27 18:51:43 -0700 (In reply to comment #25) > Before landing this, we need to figure out exactly which tests to land with it; > at the moment there are lots of different tests attached to the bug and they > need to be put in layout test form. For my future reference, what is "layout test form"? Is it documented somewhere I can find it so that I can submit future tests using it? Thanks! 13399 30 2651 oliver 2005-06-27 19:01:01 -0700 Comment on attachment 2651 edge case test for patch obsoleted by layout tests 13400 31 2652 oliver 2005-06-27 19:02:19 -0700 Comment on attachment 2652 Another edge case, slightly less ugly than previous obsoleted by layout tests 13402 32 2669 oliver 2005-06-27 19:04:22 -0700 Comment on attachment 2669 testcase for tostring with non-cyclic, but deep hierarchies moved to bug 3743 13403 33 oliver 2005-06-27 19:24:32 -0700 attachment 2665 and attachment 2666 seem to match the layout format -- at least they fit in with the the standard webkit tests 13415 34 2683 oliver 2005-06-28 03:51:04 -0700 Created attachment 2683 Quick test for join and toString operating concurrently Here's a test that demonstrates the use of join and toString in a big unholy mess. Not a webkit test atm, but should be sometime early tomorrow (NZST). toString just falls through to the join method in an Array so given the absence of concurrent execution in kjs they are fine anyway. Each call to join/toString increases the stack depth, and we currently allocate enough space to fit the entire maximum call depth. Darin: re: the hashmap should i just switch over to using that now? rather than applying the current array driven look up? and what should i be using as a hashmap implementation, from irc it sounds like there's an internal one we should be using... 15177 35 ggaren 2005-07-25 14:13:21 -0700 I have landed the patch for bug #3991. 15180 36 david 2005-07-25 14:41:04 -0700 (In reply to comment #35) > I have landed the patch for bug #3991. Uh, does that mean that this issue has been fixed? 23293 37 4532 oliver 2005-10-29 21:52:39 -0700 Created attachment 4532 Reimplemented patch with HashSet Okay, now HashSet is available in JSC I've got round to rewriting patch to use it. Needs to be reindented, and needs a half-decent testcase, but it's here now :) 23846 38 4532 darin 2005-11-09 07:39:33 -0800 Comment on attachment 4532 Reimplemented patch with HashSet I don't understand why it's necessary to copy thisObj into a second local variable currentRef. Lets just use thisObj directly unless there's a reason not to. I think it would be easier to read this if the code said: if (visitedElems.contains(thisObj)) return String(""); before anything else (even before declaring separator and str), rather than indenting all the code inside an if statement. That would also probably make the diff smaller and easier to verify. Also, I think the HashSet should be: HashSet< ObjectImp*, PointerHash<ObjectImp*> > There's an advantage to HashSet<void*>, in that we're guaranteed all such sets are a single template expansion, preventing any possible code bloat. But a disadvantage to HashSet<void*> is that without type safety, it's possible to make certain mistakes with a set (for example, adding and removing the same object, but a pointer to a different part in the case of multiple inheritance). In this case, I think the code is simple enough that I don't think there are any real problems, but I'd prefer to get off on the right foot using this class. There are a few minor formatting problems, for example: ObjectImp *currentRef=thisObj; the above needs spaces around the "=". And: + if(!visitedElems.contains(currentRef)) { the above needs a space after the if and before the opening parenthesis. And: + if ( exec->hadException() ) + break; if you're going to modify the code, you should fix the format of that by removing the extra spaces inside the parentheses. Also, there are a number of 1-line if and else statements with braces; style guide says not to use braces in these cases. 25167 39 4860 oliver 2005-11-29 17:07:36 -0800 Created attachment 4860 Hopefully the final patch Reformatted and implemented darins suggestion 25168 40 4861 oliver 2005-11-29 17:25:37 -0800 Created attachment 4861 Fixed HashMap to be typesafe 25169 41 4862 oliver 2005-11-29 17:49:30 -0800 Created attachment 4862 Testcase for toString on arrays with self references New testcase 25172 42 4863 oliver 2005-11-29 18:28:38 -0800 Created attachment 4863 Testcase, though it needs a better name 25173 43 4864 oliver 2005-11-29 18:29:29 -0800 Created attachment 4864 Expected output from testcase 25261 44 4861 mjs 2005-11-30 23:45:57 -0800 Comment on attachment 4861 Fixed HashMap to be typesafe r=me 25262 45 4861 mjs 2005-11-30 23:45:58 -0800 Comment on attachment 4861 Fixed HashMap to be typesafe r=me 2348 2005-06-14 23:27:28 -0700 2005-11-29 17:49:30 -0800 testcase testcase-javascript-circular-reference.html text/html 264 joost PGh0bWw+CjxoZWFkPgoJPHRpdGxlPlRlc3RjYXNlIGZvciBiZyAzNTM3PC90aXRsZT4KPC9oZWFk PgoKPGJvZHk+CjxzY3JpcHQgdHlwZT0idGV4dC9qYXZhc2NyaXB0Ij4KICB2YXIgYXJ5ID0gWzEs IDJdOwogIGFyeS5wdXNoKGFyeSk7CiAgZm9yICh2YXIgaSA9IDA7IGkgPCBhcnkubGVuZ3RoOyBp KyspIHsKICAgIGRvY3VtZW50LndyaXRlKHR5cGVvZiBhcnlbMF0gKyAnOiAnICsgYXJ5W2ldICsg JyAnKTsKICB9Cjwvc2NyaXB0Pgo8L2JvZHk+CjwvaHRtbD4K 2640 2005-06-25 06:32:11 -0700 2005-06-27 18:44:40 -0700 Patch to fix the *type* of error given to what it should be bug3539patch text/plain 641 oliver SW5kZXg6IC4vSmF2YVNjcmlwdENvcmUva2pzL29iamVjdC5jcHAKPT09PT09PT09PT09PT09PT09 PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PQpSQ1MgZmls ZTogL2N2cy9yb290L0phdmFTY3JpcHRDb3JlL2tqcy9vYmplY3QuY3BwLHYKcmV0cmlldmluZyBy ZXZpc2lvbiAxLjQzCmRpZmYgLXAgLXUgLXIxLjQzIC4vSmF2YVNjcmlwdENvcmUva2pzL29iamVj dC5jcHAKLS0tIC4vSmF2YVNjcmlwdENvcmUva2pzL29iamVjdC5jcHAJMjAwNS8wNi8yMiAwMDo0 ODo0NwkxLjQzCisrKyAuL0phdmFTY3JpcHRDb3JlL2tqcy9vYmplY3QuY3BwCTIwMDUvMDYvMjUg MTM6MTg6MzIKQEAgLTM4NCw3ICszODQsMTIgQEAgVmFsdWUgT2JqZWN0SW1wOjpkZWZhdWx0VmFs dWUoRXhlY1N0YXRlIAogICAgICAgfQogICAgIH0KICAgfQotCisgIAorICBpZihleGVjLT5oYWRF eGNlcHRpb24oKSl7CisgICAgICByZXR1cm4gZXhlYy0+ZXhjZXB0aW9uKCk7CisgIH0KKyAgCisg ICAgICAKICAgT2JqZWN0IGVyciA9IEVycm9yOjpjcmVhdGUoZXhlYywgVHlwZUVycm9yLCBJMThO X05PT1AoIk5vIGRlZmF1bHQgdmFsdWUiKSk7CiAgIGV4ZWMtPnNldEV4Y2VwdGlvbihlcnIpOwog ICByZXR1cm4gZXJyOwo= 2649 2005-06-25 23:21:38 -0700 2005-10-29 21:52:39 -0700 Patch to match behaviour of firefox and IE bug3539patch2 text/plain 2418 oliver SW5kZXg6IC4vSmF2YVNjcmlwdENvcmUva2pzL2FycmF5X29iamVjdC5jcHAKPT09PT09PT09PT09 PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PQpS Q1MgZmlsZTogL2N2cy9yb290L0phdmFTY3JpcHRDb3JlL2tqcy9hcnJheV9vYmplY3QuY3BwLHYK cmV0cmlldmluZyByZXZpc2lvbiAxLjQxCmRpZmYgLXAgLXUgLXIxLjQxIC4vSmF2YVNjcmlwdENv cmUva2pzL2FycmF5X29iamVjdC5jcHAKLS0tIC4vSmF2YVNjcmlwdENvcmUva2pzL2FycmF5X29i amVjdC5jcHAJMjAwNS8wNi8yMiAwMDo0ODo0NwkxLjQxCisrKyAuL0phdmFTY3JpcHRDb3JlL2tq cy9hcnJheV9vYmplY3QuY3BwCTIwMDUvMDYvMjYgMDY6MTA6MzcKQEAgLTQ0MSwxNCArNDQxLDE0 IEBAIGJvb2wgQXJyYXlQcm90b0Z1bmNJbXA6OmltcGxlbWVudHNDYWxsKCkKIFZhbHVlIEFycmF5 UHJvdG9GdW5jSW1wOjpjYWxsKEV4ZWNTdGF0ZSAqZXhlYywgT2JqZWN0ICZ0aGlzT2JqLCBjb25z dCBMaXN0ICZhcmdzKQogewogICB1bnNpZ25lZCBpbnQgbGVuZ3RoID0gdGhpc09iai5nZXQoZXhl YyxsZW5ndGhQcm9wZXJ0eU5hbWUpLnRvVUludDMyKGV4ZWMpOwotCisgIAogICBWYWx1ZSByZXN1 bHQ7CiAgIHN3aXRjaCAoaWQpIHsKICAgY2FzZSBUb0xvY2FsZVN0cmluZzoKICAgICAvLyBUT0RP ICAtIHNlZSAxNS40LjQuMwogICAgIC8vIGZhbGwgdGhyb3VnaAogICBjYXNlIFRvU3RyaW5nOgot CisgICAgCiAgICAgaWYgKCF0aGlzT2JqLmluaGVyaXRzKCZBcnJheUluc3RhbmNlSW1wOjppbmZv KSkgewogICAgICAgT2JqZWN0IGVyciA9IEVycm9yOjpjcmVhdGUoZXhlYyxUeXBlRXJyb3IpOwog ICAgICAgZXhlYy0+c2V0RXhjZXB0aW9uKGVycik7CkBAIC00NTcsMjAgKzQ1NywzNCBAQCBWYWx1 ZSBBcnJheVByb3RvRnVuY0ltcDo6Y2FsbChFeGVjU3RhdGUgCiAKICAgICAvLyBmYWxsIHRocm91 Z2gKIAotICBjYXNlIEpvaW46IHsKKyAgY2FzZSBKb2luOiB7ICAgICAKICAgICBVU3RyaW5nIHNl cGFyYXRvciA9ICIsIjsKICAgICBVU3RyaW5nIHN0ciA9ICIiOwogCi0gICAgaWYgKGFyZ3NbMF0u dHlwZSgpICE9IFVuZGVmaW5lZFR5cGUpCi0gICAgICBzZXBhcmF0b3IgPSBhcmdzWzBdLnRvU3Ry aW5nKGV4ZWMpOwotICAgIGZvciAodW5zaWduZWQgaW50IGsgPSAwOyBrIDwgbGVuZ3RoOyBrKysp IHsKLSAgICAgIGlmIChrID49IDEpCi0gICAgICAgIHN0ciArPSBzZXBhcmF0b3I7Ci0gICAgICBW YWx1ZSBlbGVtZW50ID0gdGhpc09iai5nZXQoZXhlYyxrKTsKLSAgICAgIGlmIChlbGVtZW50LnR5 cGUoKSAhPSBVbmRlZmluZWRUeXBlICYmIGVsZW1lbnQudHlwZSgpICE9IE51bGxUeXBlKQotICAg ICAgICBzdHIgKz0gZWxlbWVudC50b1N0cmluZyhleGVjKTsKLSAgICAgIGlmICggZXhlYy0+aGFk RXhjZXB0aW9uKCkgKQotCWJyZWFrOworICAgIC8vdWdseSBmaXggdG8gcHJldmVudCBpbmZpbml0 ZSByZWN1cnNpb24gaW4gdG9TdHJpbmcvam9pbgorICAgIC8vbmVjY2Vzc2FyeSB0byBtYXRjaCBi ZWhhdmlvdXIgb2YgZmlyZWZveCBhbmQgSUUKKyAgICBzdGF0aWMgT2JqZWN0SW1wKiB2aXNpdGVk UmVmc1tLSlNfTUFYX1NUQUNLXTsgCisgICAgc3RhdGljIGludCBqb2luRGVwdGg9MDsKKyAgICBP YmplY3RJbXAgKmN1cnJlbnRSZWY9dGhpc09iai5pbXAoKTsKKyAgICBpbnQgbWF0Y2hOdW09MDsK KyAgICBmb3IobWF0Y2hOdW09MDsgbWF0Y2hOdW08am9pbkRlcHRoJiYoY3VycmVudFJlZiE9dmlz aXRlZFJlZnNbbWF0Y2hOdW1dKTsgbWF0Y2hOdW0rKyl7CisgICAgfQorICAgIAorICAgIGlmKG1h dGNoTnVtPT1qb2luRGVwdGgpeworICAgICAgLy93ZSBoYXZlbid0IGFscmVhZHkgb3V0cHV0IHRo aXMgYXJyYXkKKyAgICAgIHZpc2l0ZWRSZWZzW2pvaW5EZXB0aCsrXT1jdXJyZW50UmVmOworICAg ICAgaWYgKGFyZ3NbMF0udHlwZSgpICE9IFVuZGVmaW5lZFR5cGUpCisgICAgICAgIHNlcGFyYXRv ciA9IGFyZ3NbMF0udG9TdHJpbmcoZXhlYyk7CisgICAgICBmb3IgKHVuc2lnbmVkIGludCBrID0g MDsgayA8IGxlbmd0aDsgaysrKSB7CisgICAgICAgIGlmIChrID49IDEpCisgICAgICAgICAgc3Ry ICs9IHNlcGFyYXRvcjsKKyAgICAgICAgVmFsdWUgZWxlbWVudCA9IHRoaXNPYmouZ2V0KGV4ZWMs ayk7CisgICAgICAgIGlmIChlbGVtZW50LnR5cGUoKSAhPSBVbmRlZmluZWRUeXBlICYmIGVsZW1l bnQudHlwZSgpICE9IE51bGxUeXBlKQorICAgICAgICAgIHN0ciArPSBlbGVtZW50LnRvU3RyaW5n KGV4ZWMpOworICAgICAgICBpZiAoIGV4ZWMtPmhhZEV4Y2VwdGlvbigpICkKKyAgICAgICAgICBi cmVhazsKKyAgICAgIH0KKyAgICAgIGpvaW5EZXB0aC0tOwogICAgIH0KICAgICByZXN1bHQgPSBT dHJpbmcoc3RyKTsKICAgICBicmVhazsK 2651 2005-06-26 00:19:43 -0700 2005-06-27 19:01:01 -0700 edge case test for patch Testcase for bg 3537.html text/html 295 oliver PGh0bWw+CjxoZWFkPgoJPHRpdGxlPlRlc3RjYXNlIGZvciBKUyBiYWRuZXNzPC90aXRsZT4KPC9o ZWFkPgoKPGJvZHk+CjxzY3JpcHQgdHlwZT0idGV4dC9qYXZhc2NyaXB0Ij4KICB2YXIgZWxlbT1b XTsKICBmb3IodmFyIGk9MDsgaTwxMDAwOyBpKyspCiAgewogICAgZWxlbS5wdXNoKFtpXSk7CiAg fQogIGZvcih2YXIgaT0wOyBpPDEwMDA7IGkrKykKICB7CiAgICBlbGVtW2ldLnB1c2goZWxlbVso aSsxKSUxMDAwXSk7CiAgfQogIGRvY3VtZW50LndyaXRlKGVsZW0pOwo8L3NjcmlwdD4KPC9ib2R5 Pgo8L2h0bWw+Cg== 2652 2005-06-26 01:12:40 -0700 2005-06-27 19:02:19 -0700 Another edge case, slightly less ugly than previous Testcase for bg 3537.html text/html 243 oliver PGh0bWw+CjxoZWFkPgoJPHRpdGxlPlRlc3RjYXNlIGZvciBKUyBiYWRuZXNzPC90aXRsZT4KPC9o ZWFkPgoKPGJvZHk+CjxzY3JpcHQgdHlwZT0idGV4dC9qYXZhc2NyaXB0Ij4KICB2YXIgZWxlbT1b MF07CiAgZm9yKHZhciBpPTE7IGk8PTEwMDAwOyBpKyspCiAgewogICAgdmFyIGFyeT1bZWxlbSwg aV07CiAgICBlbGVtPWFyeTsKICB9CiAgZG9jdW1lbnQud3JpdGUoZWxlbSk7Cjwvc2NyaXB0Pgo8 L2JvZHk+CjwvaHRtbD4K 2665 2005-06-27 04:54:20 -0700 2005-11-29 17:49:30 -0800 test case for cyclic references in array.toString cyclic-references.html text/html 952 oliver PCFET0NUWVBFIEhUTUwgUFVCTElDICItLy9XM0MvL0RURCBIVE1MIDQuMDEvL0VOIiAiaHR0cDov L3d3dy53My5vcmcvVFIvaHRtbDQvc3RyaWN0LmR0ZCI+CjxodG1sPgo8aGVhZD4KPHRpdGxlPnRl c3QgY3ljbGljIHJlZmVyZW5jZXMgYW5kIEFycmF5LnRvU3RyaW5nPC90aXRsZT4KPHNjcmlwdCBs YW5ndWFnZT0iamF2YXNjcmlwdCI+CmZ1bmN0aW9uIGluaXQoKXsKICB2YXIgYXJ5MT1bMSwyXTsK ICBhcnkxLnB1c2goYXJ5MSk7CiAgZG9jdW1lbnQuZ2V0RWxlbWVudEJ5SWQoInIwYzAiKS5maXJz dENoaWxkLm5vZGVWYWx1ZSA9IGFyeTE7CiAgYXJ5MT1bMSwyXTsKICB2YXIgYXJ5Mj1bMyw0XTsK ICBhcnkxLnB1c2goYXJ5Mik7CiAgYXJ5Mi5wdXNoKGFyeTEpOwogIGRvY3VtZW50LmdldEVsZW1l bnRCeUlkKCJyMWMwIikuZmlyc3RDaGlsZC5ub2RlVmFsdWUgPSBhcnkxOwogIGFyeTEucHVzaCg1 KTsKICBkb2N1bWVudC5nZXRFbGVtZW50QnlJZCgicjJjMCIpLmZpcnN0Q2hpbGQubm9kZVZhbHVl ID0gYXJ5MTsKICBpZiAod2luZG93LmxheW91dFRlc3RDb250cm9sbGVyKSB7CiAgICB3aW5kb3cu bGF5b3V0VGVzdENvbnRyb2xsZXIuZHVtcEFzVGV4dCgpOwogIH0KfQo8L3NjcmlwdD4KPC9oZWFk Pgo8Ym9keSBvbmxvYWQ9ImluaXQoKSI+CiAgICBUaGlzIHdpbGwgdGVzdCBBcnJheS50b1N0cmlu ZyB3aXRoIGN5Y2xpYyByZWZlcmVuY2VzLgogICAgPHRhYmxlIGNlbGxwYWRkaW5nPSIyIiBjZWxs c3BhY2luZz0iMiIgYm9yZGVyPSIxIj4KCQk8dGJvZHk+CgkJCTx0cj4KCQkJCTx0ZCBpZD0icjBj MCI+Jm5ic3A7PC90ZD4KCQkJPC90cj4KCQkJPHRyPgoJCQkJPHRkIGlkPSJyMWMwIj4mbmJzcDs8 L3RkPgoJCQk8L3RyPgoJCQk8dHI+CgkJCQk8dGQgaWQ9InIyYzAiPiZuYnNwOzwvdGQ+CgkJCTwv dHI+CgkJPC90Ym9keT4KCTwvdGFibGU+CjwvYm9keT4KPC9odG1sPg== 2666 2005-06-27 04:56:11 -0700 2005-11-29 17:49:30 -0800 expected results of cyclic reference testcase cyclic-references-expected.txt text/plain 80 oliver VGhpcyB3aWxsIHRlc3QgQXJyYXkudG9TdHJpbmcgd2l0aCBjeWNsaWMgcmVmZXJlbmNlcy4KMSwy LAoxLDIsMyw0LAoxLDIsMyw0LCw1Cgo= 2669 2005-06-27 05:30:22 -0700 2005-06-27 19:04:22 -0700 testcase for tostring with non-cyclic, but deep hierarchies tostring-deep-hierarchy.html text/html 925 oliver PCFET0NUWVBFIEhUTUwgUFVCTElDICItLy9XM0MvL0RURCBIVE1MIDQuMDEvL0VOIiAiaHR0cDov L3d3dy53My5vcmcvVFIvaHRtbDQvc3RyaWN0LmR0ZCI+CjxodG1sPgo8aGVhZD4KPHRpdGxlPlRl c3RpbmcgZGVlcCBoaWVyYXJjaGllcyB3aXRoIEFycmF5LnRvU3RyaW5nPC90aXRsZT4KPHNjcmlw dCBsYW5ndWFnZT0iamF2YXNjcmlwdCI+CmZ1bmN0aW9uIGluaXQoKXsKICB2YXIgYXJ5MT1bMF07 CiAgZm9yKHZhciBpPTE7IGk8MTA7IGkrKykKICAgIGFyeTE9W2FyeTEsIGldCiAgZG9jdW1lbnQu Z2V0RWxlbWVudEJ5SWQoInIwYzAiKS5maXJzdENoaWxkLm5vZGVWYWx1ZSA9IGFyeTE7CiAgCiAg dmFyIGFyeTI9WzBdOwogIGZvcih2YXIgaT0xOyBpPDEwMDAwOyBpKyspCiAgICBhcnkyPVthcnky LCBpXTsKICAgIAogIHZhciBvdXQ7CiAgdHJ5ewogICAgb3V0PWFyeTIudG9TdHJpbmcoKTsKICB9 Y2F0Y2goZSl7CiAgICBvdXQ9ZTsKICB9CiAgZG9jdW1lbnQuZ2V0RWxlbWVudEJ5SWQoInIxYzAi KS5maXJzdENoaWxkLm5vZGVWYWx1ZSA9IG91dDsKICBpZiAod2luZG93LmxheW91dFRlc3RDb250 cm9sbGVyKSB7CiAgICB3aW5kb3cubGF5b3V0VGVzdENvbnRyb2xsZXIuZHVtcEFzVGV4dCgpOwog IH0KfQo8L3NjcmlwdD4KPC9oZWFkPgo8Ym9keSBvbmxvYWQ9ImluaXQoKSI+CiAgICBUaGlzIHdp bGwgdGVzdCBBcnJheS50b1N0cmluZyB3aXRoIGRlZXAgaGllcmFyY2hpZXMuCiAgICA8dGFibGUg Y2VsbHBhZGRpbmc9IjIiIGNlbGxzcGFjaW5nPSIyIiBib3JkZXI9IjEiPgoJCTx0Ym9keT4KCQkJ PHRyPgoJCQkJPHRkIGlkPSJyMGMwIj4mbmJzcDs8L3RkPgoJCQk8L3RyPgoJCQk8dHI+CgkJCQk8 dGQgaWQ9InIxYzAiPiZuYnNwOzwvdGQ+CgkJCTwvdHI+CgkJPC90Ym9keT4KCTwvdGFibGU+Cjwv Ym9keT4KPC9odG1sPg== 2670 2005-06-27 05:31:38 -0700 2005-06-27 18:41:33 -0700 expected output from the deep hierarchies test tostring-deep-hierarchy-expected.txt text/plain 121 oliver VGhpcyB3aWxsIHRlc3QgQXJyYXkudG9TdHJpbmcgd2l0aCBkZWVwIGhpZXJhcmNoaWVzLgowLDEs MiwzLDQsNSw2LDcsOCw5ClJhbmdlRXJyb3IgLSBNYXhpbXVtIGNhbGwgc3RhY2sgc2l6ZSBleGNl ZWRlZC4KCg== 2676 2005-06-27 15:45:45 -0700 2005-11-29 17:49:30 -0800 sort() Test Case try.html text/html 286 david PGh0bWw+CjxoZWFkPgoJPHRpdGxlPlRlc3RjYXNlIGZvciBiZyAzNTM3IChzb3J0KCkpPC90aXRs ZT4KPC9oZWFkPgoKPGJvZHk+CjxzY3JpcHQgdHlwZT0idGV4dC9qYXZhc2NyaXB0Ij4KICB2YXIg YXJ5ID0gWzEsIDJdOwogIGFyeS5wdXNoKGFyeSk7CiAgYXJ5LnNvcnQoKTsgLy8gQ29tbWVudCBv dXQgdGhpcyBsaW5lIHRvIGdldCB3b3JraW5nIGFnYWluLgogIGRvY3VtZW50LndyaXRlKCJ0aGlz IHNob3VsZCBiZSBvdXRwdXQsIGJ1dCBpc24ndCIpOwo8L3NjcmlwdD4KPC9ib2R5Pgo8L2h0bWw+ Cg== 2683 2005-06-28 03:51:04 -0700 2005-11-29 17:49:30 -0800 Quick test for join and toString operating concurrently jsjointest.html text/html 414 oliver PGh0bWw+CjxoZWFkPgoJPHRpdGxlPlRlc3RjYXNlIGZvciBiZyAzNTM5PC90aXRsZT4KPC9oZWFk PgoKPGJvZHk+CjxzY3JpcHQgdHlwZT0idGV4dC9qYXZhc2NyaXB0Ij4KICBmdW5jdGlvbiBBcnlS ZWYoYXJ5LCBpKXsKICAgIHRoaXMuYXJ5PVtpLCBhcnksIGldOwogICAgdGhpcy50b1N0cmluZz10 ZXN0VG9TdHJpbmc7CiAgfQogIGZ1bmN0aW9uIHRlc3RUb1N0cmluZygpewogICAgcmV0dXJuICIo Iit0aGlzLmFyeS5qb2luKCJ8IikrIikiOwogIH0KICB2YXIgYXJ5PVtdOwogIHZhciBhPW5ldyBB cnlSZWYoYXJ5KTsKICBhcnkucHVzaChhKTsKICBmb3IodmFyIGk9MTsgaTwxMDsgaSsrKXsKICAg IGE9bmV3IEFyeVJlZihhLCBpKTsKICB9CiAgZG9jdW1lbnQud3JpdGUoYSk7Cjwvc2NyaXB0Pgo8 L2JvZHk+CjwvaHRtbD4K 4532 2005-10-29 21:52:39 -0700 2005-11-29 17:07:36 -0800 Reimplemented patch with HashSet selfreference.patch text/plain 3740 oliver SW5kZXg6IGtqcy9hcnJheV9vYmplY3QuY3BwCj09PT09PT09PT09PT09PT09PT09PT09PT09PT09 PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT0KUkNTIGZpbGU6IC9jdnMvcm9v dC9KYXZhU2NyaXB0Q29yZS9ranMvYXJyYXlfb2JqZWN0LmNwcCx2CnJldHJpZXZpbmcgcmV2aXNp b24gMS41NApkaWZmIC1wIC11IC1yMS41NCBranMvYXJyYXlfb2JqZWN0LmNwcAotLS0ga2pzL2Fy cmF5X29iamVjdC5jcHAJMTYgT2N0IDIwMDUgMDA6NDY6MjEgLTAwMDAJMS41NAorKysga2pzL2Fy cmF5X29iamVjdC5jcHAJMzAgT2N0IDIwMDUgMDQ6NTM6MTEgLTAwMDAKQEAgLTMxLDYgKzMxLDcg QEAKICNpbmNsdWRlICJyZWZlcmVuY2VfbGlzdC5oIgogI2luY2x1ZGUgInR5cGVzLmgiCiAjaW5j bHVkZSAidmFsdWUuaCIKKyNpbmNsdWRlICJIYXNoU2V0LmgiCiAKICNpbmNsdWRlICJhcnJheV9v YmplY3QubHV0LmgiCiAKQEAgLTQ0OCw0NSArNDQ5LDUwIEBAIFZhbHVlSW1wICpBcnJheVByb3Rv RnVuY0ltcDo6Y2FsbEFzRnVuY3QKICAgY2FzZSBKb2luOiB7CiAgICAgVVN0cmluZyBzZXBhcmF0 b3IgPSAiLCI7CiAgICAgVVN0cmluZyBzdHIgPSAiIjsKLQotICAgIGlmICghYXJnc1swXS0+aXNV bmRlZmluZWQoKSkKLSAgICAgIHNlcGFyYXRvciA9IGFyZ3NbMF0tPnRvU3RyaW5nKGV4ZWMpOwot ICAgIGZvciAodW5zaWduZWQgaW50IGsgPSAwOyBrIDwgbGVuZ3RoOyBrKyspIHsKLSAgICAgIGlm IChrID49IDEpCi0gICAgICAgIHN0ciArPSBzZXBhcmF0b3I7Ci0gICAgICAKLSAgICAgIFZhbHVl SW1wICplbGVtZW50ID0gdGhpc09iai0+Z2V0KGV4ZWMsIGspOwotICAgICAgaWYgKGVsZW1lbnQt PmlzVW5kZWZpbmVkT3JOdWxsKCkpCi0gICAgICAgIGNvbnRpbnVlOwotCi0gICAgICBib29sIGZh bGxiYWNrID0gZmFsc2U7Ci0gICAgICBpZiAoaWQgPT0gVG9Mb2NhbGVTdHJpbmcpIHsKLSAgICAg ICAgT2JqZWN0SW1wICpvID0gZWxlbWVudC0+dG9PYmplY3QoZXhlYyk7Ci0gICAgICAgIFZhbHVl SW1wICpjb252ZXJzaW9uRnVuY3Rpb24gPSBvLT5nZXQoZXhlYywgdG9Mb2NhbGVTdHJpbmdQcm9w ZXJ0eU5hbWUpOwotICAgICAgICBpZiAoY29udmVyc2lvbkZ1bmN0aW9uLT5pc09iamVjdCgpICYm IHN0YXRpY19jYXN0PE9iamVjdEltcCAqPihjb252ZXJzaW9uRnVuY3Rpb24pLT5pbXBsZW1lbnRz Q2FsbCgpKSB7Ci0gICAgICAgICAgc3RyICs9IHN0YXRpY19jYXN0PE9iamVjdEltcCAqPihjb252 ZXJzaW9uRnVuY3Rpb24pLT5jYWxsKGV4ZWMsIG8sIExpc3QoKSktPnRvU3RyaW5nKGV4ZWMpOwot ICAgICAgICB9IGVsc2UgewotICAgICAgICAgIC8vIHRyeSB0b1N0cmluZygpIGZhbGxiYWNrCi0g ICAgICAgICAgZmFsbGJhY2sgPSB0cnVlOwotICAgICAgICB9Ci0gICAgICB9Ci0KLSAgICAgIGlm IChpZCA9PSBUb1N0cmluZyB8fCBpZCA9PSBKb2luIHx8IGZhbGxiYWNrKSB7Ci0gICAgICAgIGlm IChlbGVtZW50LT5pc09iamVjdCgpKSB7Ci0gICAgICAgICAgT2JqZWN0SW1wICpvID0gc3RhdGlj X2Nhc3Q8T2JqZWN0SW1wICo+KGVsZW1lbnQpOwotICAgICAgICAgIFZhbHVlSW1wICpjb252ZXJz aW9uRnVuY3Rpb24gPSBvLT5nZXQoZXhlYywgdG9TdHJpbmdQcm9wZXJ0eU5hbWUpOworICAgIHN0 YXRpYyBIYXNoU2V0PHZvaWQqPiB2aXNpdGVkRWxlbXM7CisgICAgT2JqZWN0SW1wICpjdXJyZW50 UmVmPXRoaXNPYmo7CisgICAgaWYoIXZpc2l0ZWRFbGVtcy5jb250YWlucyhjdXJyZW50UmVmKSkg eyAgICAKKyAgICAgIHZpc2l0ZWRFbGVtcy5pbnNlcnQoY3VycmVudFJlZik7CisgICAgICBpZiAo IWFyZ3NbMF0tPmlzVW5kZWZpbmVkKCkpCisgICAgICAgIHNlcGFyYXRvciA9IGFyZ3NbMF0tPnRv U3RyaW5nKGV4ZWMpOworICAgICAgZm9yICh1bnNpZ25lZCBpbnQgayA9IDA7IGsgPCBsZW5ndGg7 IGsrKykgeworICAgICAgICBpZiAoayA+PSAxKQorICAgICAgICAgIHN0ciArPSBzZXBhcmF0b3I7 CisgICAgICAgCisgICAgICAgIFZhbHVlSW1wICplbGVtZW50ID0gdGhpc09iai0+Z2V0KGV4ZWMs IGspOworICAgICAgICBpZiAoZWxlbWVudC0+aXNVbmRlZmluZWRPck51bGwoKSkKKyAgICAgICAg ICBjb250aW51ZTsKKworICAgICAgICBib29sIGZhbGxiYWNrID0gZmFsc2U7CisgICAgICAgIGlm IChpZCA9PSBUb0xvY2FsZVN0cmluZykgeworICAgICAgICAgIE9iamVjdEltcCAqbyA9IGVsZW1l bnQtPnRvT2JqZWN0KGV4ZWMpOworICAgICAgICAgIFZhbHVlSW1wICpjb252ZXJzaW9uRnVuY3Rp b24gPSBvLT5nZXQoZXhlYywgdG9Mb2NhbGVTdHJpbmdQcm9wZXJ0eU5hbWUpOwogICAgICAgICAg IGlmIChjb252ZXJzaW9uRnVuY3Rpb24tPmlzT2JqZWN0KCkgJiYgc3RhdGljX2Nhc3Q8T2JqZWN0 SW1wICo+KGNvbnZlcnNpb25GdW5jdGlvbiktPmltcGxlbWVudHNDYWxsKCkpIHsKICAgICAgICAg ICAgIHN0ciArPSBzdGF0aWNfY2FzdDxPYmplY3RJbXAgKj4oY29udmVyc2lvbkZ1bmN0aW9uKS0+ Y2FsbChleGVjLCBvLCBMaXN0KCkpLT50b1N0cmluZyhleGVjKTsKICAgICAgICAgICB9IGVsc2Ug ewotICAgICAgICAgICAgcmV0dXJuIHRocm93RXJyb3IoZXhlYywgUmFuZ2VFcnJvciwgIkNhbid0 IGNvbnZlcnQgIiArIG8tPmNsYXNzTmFtZSgpICsgIiBvYmplY3QgdG8gc3RyaW5nIik7CisgICAg ICAgICAgICAvLyB0cnkgdG9TdHJpbmcoKSBmYWxsYmFjaworICAgICAgICAgICAgZmFsbGJhY2sg PSB0cnVlOwogICAgICAgICAgIH0KLSAgICAgICAgfSBlbHNlIHsKLSAgICAgICAgICBzdHIgKz0g ZWxlbWVudC0+dG9TdHJpbmcoZXhlYyk7CiAgICAgICAgIH0KLSAgICAgIH0KIAotICAgICAgaWYg KCBleGVjLT5oYWRFeGNlcHRpb24oKSApCi0gICAgICAgIGJyZWFrOworICAgICAgICBpZiAoaWQg PT0gVG9TdHJpbmcgfHwgaWQgPT0gSm9pbiB8fCBmYWxsYmFjaykgeworICAgICAgICAgIGlmIChl bGVtZW50LT5pc09iamVjdCgpKSB7CisgICAgICAgICAgICBPYmplY3RJbXAgKm8gPSBzdGF0aWNf Y2FzdDxPYmplY3RJbXAgKj4oZWxlbWVudCk7CisgICAgICAgICAgICBWYWx1ZUltcCAqY29udmVy c2lvbkZ1bmN0aW9uID0gby0+Z2V0KGV4ZWMsIHRvU3RyaW5nUHJvcGVydHlOYW1lKTsKKyAgICAg ICAgICAgIGlmIChjb252ZXJzaW9uRnVuY3Rpb24tPmlzT2JqZWN0KCkgJiYgc3RhdGljX2Nhc3Q8 T2JqZWN0SW1wICo+KGNvbnZlcnNpb25GdW5jdGlvbiktPmltcGxlbWVudHNDYWxsKCkpIHsKKyAg ICAgICAgICAgICAgc3RyICs9IHN0YXRpY19jYXN0PE9iamVjdEltcCAqPihjb252ZXJzaW9uRnVu Y3Rpb24pLT5jYWxsKGV4ZWMsIG8sIExpc3QoKSktPnRvU3RyaW5nKGV4ZWMpOworICAgICAgICAg ICAgfSBlbHNlIHsKKyAgICAgICAgICAgICAgcmV0dXJuIHRocm93RXJyb3IoZXhlYywgUmFuZ2VF cnJvciwgIkNhbid0IGNvbnZlcnQgIiArIG8tPmNsYXNzTmFtZSgpICsgIiBvYmplY3QgdG8gc3Ry aW5nIik7CisgICAgICAgICAgICB9CisgICAgICAgICAgfSBlbHNlIHsKKyAgICAgICAgICAgIHN0 ciArPSBlbGVtZW50LT50b1N0cmluZyhleGVjKTsKKyAgICAgICAgICB9CisgICAgICAgIH0KKwor ICAgICAgICBpZiAoIGV4ZWMtPmhhZEV4Y2VwdGlvbigpICkKKyAgICAgICAgICBicmVhazsKKyAg ICAgIH0KKyAgICAgIHZpc2l0ZWRFbGVtcy5yZW1vdmUoY3VycmVudFJlZik7CiAgICAgfQogICAg IHJlc3VsdCA9IFN0cmluZyhzdHIpOwogICAgIGJyZWFrOwo= 4860 2005-11-29 17:07:36 -0800 2005-11-29 17:23:37 -0800 Hopefully the final patch circrefs.patch text/plain 1095 oliver SW5kZXg6IGtqcy9hcnJheV9vYmplY3QuY3BwCj09PT09PT09PT09PT09PT09PT09PT09PT09PT09 PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT0KUkNTIGZpbGU6IC9jdnMvcm9v dC9KYXZhU2NyaXB0Q29yZS9ranMvYXJyYXlfb2JqZWN0LmNwcCx2CnJldHJpZXZpbmcgcmV2aXNp b24gMS41NApkaWZmIC1wIC11IC1yMS41NCBranMvYXJyYXlfb2JqZWN0LmNwcAotLS0ga2pzL2Fy cmF5X29iamVjdC5jcHAJMTYgT2N0IDIwMDUgMDA6NDY6MjEgLTAwMDAJMS41NAorKysga2pzL2Fy cmF5X29iamVjdC5jcHAJMzAgTm92IDIwMDUgMDE6MTI6MDggLTAwMDAKQEAgLTMxLDYgKzMxLDcg QEAKICNpbmNsdWRlICJyZWZlcmVuY2VfbGlzdC5oIgogI2luY2x1ZGUgInR5cGVzLmgiCiAjaW5j bHVkZSAidmFsdWUuaCIKKyNpbmNsdWRlICJIYXNoU2V0LmgiCiAKICNpbmNsdWRlICJhcnJheV9v YmplY3QubHV0LmgiCiAKQEAgLTQ0Niw5ICs0NDcsMTMgQEAgVmFsdWVJbXAgKkFycmF5UHJvdG9G dW5jSW1wOjpjYWxsQXNGdW5jdAogCiAgICAgLy8gZmFsbCB0aHJvdWdoCiAgIGNhc2UgSm9pbjog eworICAgIHN0YXRpYyBIYXNoU2V0PHZvaWQqPiB2aXNpdGVkRWxlbXM7CisgICAgaWYgKHZpc2l0 ZWRFbGVtcy5jb250YWlucyh0aGlzT2JqKSkKKyAgICAgIHJldHVybiBTdHJpbmcoIiIpOwogICAg IFVTdHJpbmcgc2VwYXJhdG9yID0gIiwiOwogICAgIFVTdHJpbmcgc3RyID0gIiI7CiAKKyAgICB2 aXNpdGVkRWxlbXMuaW5zZXJ0KHRoaXNPYmopOwogICAgIGlmICghYXJnc1swXS0+aXNVbmRlZmlu ZWQoKSkKICAgICAgIHNlcGFyYXRvciA9IGFyZ3NbMF0tPnRvU3RyaW5nKGV4ZWMpOwogICAgIGZv ciAodW5zaWduZWQgaW50IGsgPSAwOyBrIDwgbGVuZ3RoOyBrKyspIHsKQEAgLTQ4OCw2ICs0OTMs NyBAQCBWYWx1ZUltcCAqQXJyYXlQcm90b0Z1bmNJbXA6OmNhbGxBc0Z1bmN0CiAgICAgICBpZiAo IGV4ZWMtPmhhZEV4Y2VwdGlvbigpICkKICAgICAgICAgYnJlYWs7CiAgICAgfQorICAgIHZpc2l0 ZWRFbGVtcy5yZW1vdmUodGhpc09iaik7CiAgICAgcmVzdWx0ID0gU3RyaW5nKHN0cik7CiAgICAg YnJlYWs7CiAgIH0K 4861 2005-11-29 17:25:37 -0800 2005-11-30 23:45:57 -0800 Fixed HashMap to be typesafe circrefs.patch text/plain 1128 oliver SW5kZXg6IGtqcy9hcnJheV9vYmplY3QuY3BwCj09PT09PT09PT09PT09PT09PT09PT09PT09PT09 PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT0KUkNTIGZpbGU6IC9jdnMvcm9v dC9KYXZhU2NyaXB0Q29yZS9ranMvYXJyYXlfb2JqZWN0LmNwcCx2CnJldHJpZXZpbmcgcmV2aXNp b24gMS41NApkaWZmIC1wIC11IC1yMS41NCBranMvYXJyYXlfb2JqZWN0LmNwcAotLS0ga2pzL2Fy cmF5X29iamVjdC5jcHAJMTYgT2N0IDIwMDUgMDA6NDY6MjEgLTAwMDAJMS41NAorKysga2pzL2Fy cmF5X29iamVjdC5jcHAJMzAgTm92IDIwMDUgMDE6Mjk6MzEgLTAwMDAKQEAgLTMxLDYgKzMxLDcg QEAKICNpbmNsdWRlICJyZWZlcmVuY2VfbGlzdC5oIgogI2luY2x1ZGUgInR5cGVzLmgiCiAjaW5j bHVkZSAidmFsdWUuaCIKKyNpbmNsdWRlICJIYXNoU2V0LmgiCiAKICNpbmNsdWRlICJhcnJheV9v YmplY3QubHV0LmgiCiAKQEAgLTQ0Niw5ICs0NDcsMTMgQEAgVmFsdWVJbXAgKkFycmF5UHJvdG9G dW5jSW1wOjpjYWxsQXNGdW5jdAogCiAgICAgLy8gZmFsbCB0aHJvdWdoCiAgIGNhc2UgSm9pbjog eworICAgIHN0YXRpYyBIYXNoU2V0PE9iamVjdEltcCAqLCBQb2ludGVySGFzaDxPYmplY3RJbXAg Kj4gPiB2aXNpdGVkRWxlbXM7CisgICAgaWYgKHZpc2l0ZWRFbGVtcy5jb250YWlucyh0aGlzT2Jq KSkKKyAgICAgIHJldHVybiBTdHJpbmcoIiIpOwogICAgIFVTdHJpbmcgc2VwYXJhdG9yID0gIiwi OwogICAgIFVTdHJpbmcgc3RyID0gIiI7CiAKKyAgICB2aXNpdGVkRWxlbXMuaW5zZXJ0KHRoaXNP YmopOwogICAgIGlmICghYXJnc1swXS0+aXNVbmRlZmluZWQoKSkKICAgICAgIHNlcGFyYXRvciA9 IGFyZ3NbMF0tPnRvU3RyaW5nKGV4ZWMpOwogICAgIGZvciAodW5zaWduZWQgaW50IGsgPSAwOyBr IDwgbGVuZ3RoOyBrKyspIHsKQEAgLTQ4OCw2ICs0OTMsNyBAQCBWYWx1ZUltcCAqQXJyYXlQcm90 b0Z1bmNJbXA6OmNhbGxBc0Z1bmN0CiAgICAgICBpZiAoIGV4ZWMtPmhhZEV4Y2VwdGlvbigpICkK ICAgICAgICAgYnJlYWs7CiAgICAgfQorICAgIHZpc2l0ZWRFbGVtcy5yZW1vdmUodGhpc09iaik7 CiAgICAgcmVzdWx0ID0gU3RyaW5nKHN0cik7CiAgICAgYnJlYWs7CiAgIH0K 4862 2005-11-29 17:49:30 -0800 2005-11-29 18:28:38 -0800 Testcase for toString on arrays with self references cyclicreftest.html text/html 1523 oliver PCFET0NUWVBFIEhUTUwgUFVCTElDICItLy9XM0MvL0RURCBIVE1MIDQuMDEvL0VOIiAiaHR0cDov L3d3dy53My5vcmcvVFIvaHRtbDQvc3RyaWN0LmR0ZCI+CjxodG1sPgo8aGVhZD4KPHRpdGxlPnRl c3QgY3ljbGljIHJlZmVyZW5jZXMgYW5kIEFycmF5LnRvU3RyaW5nPC90aXRsZT4KPHNjcmlwdCBs YW5ndWFnZT0iamF2YXNjcmlwdCI+CmZ1bmN0aW9uIGluaXQoKXsKICB2YXIgYXJ5MT1bMSwyXTsK ICBhcnkxLnB1c2goYXJ5MSk7CiAgZG9jdW1lbnQuZ2V0RWxlbWVudEJ5SWQoInIwYzEiKS5maXJz dENoaWxkLm5vZGVWYWx1ZSA9IGFyeTE7CiAgYXJ5MT1bMSwyXTsKICB2YXIgYXJ5Mj1bMyw0XTsK ICBhcnkxLnB1c2goYXJ5Mik7CiAgYXJ5Mi5wdXNoKGFyeTEpOwogIGRvY3VtZW50LmdldEVsZW1l bnRCeUlkKCJyMWMxIikuZmlyc3RDaGlsZC5ub2RlVmFsdWUgPSBhcnkxOwogIGFyeTEucHVzaCg1 KTsKICBkb2N1bWVudC5nZXRFbGVtZW50QnlJZCgicjJjMSIpLmZpcnN0Q2hpbGQubm9kZVZhbHVl ID0gYXJ5MTsKICBpZiAod2luZG93LmxheW91dFRlc3RDb250cm9sbGVyKSB7CiAgICB3aW5kb3cu bGF5b3V0VGVzdENvbnRyb2xsZXIuZHVtcEFzVGV4dCgpOwogIH0KfQo8L3NjcmlwdD4KPC9oZWFk Pgo8Ym9keSBvbmxvYWQ9ImluaXQoKSI+CiAgICBUaGlzIHdpbGwgdGVzdCBBcnJheS50b1N0cmlu ZyB3aXRoIGNpcmN1bGFyIHJlZmVyZW5jZXMuICBJZiBhbiBlbGVtZW50IGNvbnRhaW5zIGEgcmVm ZXJlbmNlIHRvIAogICAgaXRzZWxmIG9yIG9uZSBvZiBpdHMgY2hpbGRyZW4gKGF0IGFueSBkZXB0 aCkgY29udGFpbnMgYSByZWZlcmVuY2UgdG8gaXQsIGl0IHdpbGwgYmUgc2tpcHBlZC4KICAgIFRo aXMgY2FuIHJlc3VsdCBpbiBlaXRoZXIgYSB0cmFpbGluZyAnLCcgKGlmIHRoZSBzZWxmIHJlZmVy ZW5jZSBpcyBhdCB0aGUgZW5kIG9mIHRoZSBhcnJheSkgb3IgJywsJyAKICAgIGlmIHRoZSBzZWxm IHJlZmVyZW5jZSBpcyBjb250YWluZWQgYXQgc29tZSBtaWQgcG9pbnQgaW4gdGhlIGFycmF5Lgog ICAgPHRhYmxlIGNlbGxwYWRkaW5nPSIyIiBjZWxsc3BhY2luZz0iMiIgYm9yZGVyPSIxIj4KCQk8 dGJvZHk+CgkJCTx0cj4KCQkJCTx0ZD5BcnJheSBDb250ZW50czwvdGQ+PHRkPkV4cGVjdGVkIE91 dHB1dDwvdGQ+PHRkPkFjdHVhbCBPdXRwdXQ8L3RkPgoJCQk8L3RyPgoJCQk8dHI+CgkJCQk8dGQ+ WzEsIDIsIHNlbGZdPC90ZD48dGQgaWQ9InIwYzAiPjEsMiw8L3RkPjx0ZCBpZD0icjBjMSI+Jm5i c3A7PC90ZD4KCQkJPC90cj4KCQkJPHRyPgoJCQkJPHRkPlsxLCAyLCBbMywgNCwgcGFyZW50XV08 L3RkPjx0ZCBpZD0icjFjMCI+MSwyLDMsNCw8L3RkPjx0ZCBpZD0icjFjMSI+Jm5ic3A7PC90ZD4K CQkJPC90cj4KCQkJPHRyPgoJCQkJPHRkPlsxLCAyLCBbMywgNCwgcGFyZW50XSwgNV08L3RkPjx0 ZCBpZD0icjJjMCI+MSwyLDMsNCwsNTwvdGQ+PHRkIGlkPSJyMmMxIj4mbmJzcDs8L3RkPgoJCQk8 L3RyPgoJCTwvdGJvZHk+Cgk8L3RhYmxlPgo8L2JvZHk+CjwvaHRtbD4= 4863 2005-11-29 18:28:38 -0800 2005-11-29 18:28:38 -0800 Testcase, though it needs a better name cyclicreftest.html text/html 1538 oliver PCFET0NUWVBFIEhUTUwgUFVCTElDICItLy9XM0MvL0RURCBIVE1MIDQuMDEvL0VOIiAiaHR0cDov L3d3dy53My5vcmcvVFIvaHRtbDQvc3RyaWN0LmR0ZCI+CjxodG1sPgo8aGVhZD4KPHRpdGxlPnRl c3QgY3ljbGljIHJlZmVyZW5jZXMgYW5kIEFycmF5LnRvU3RyaW5nPC90aXRsZT4KPHNjcmlwdCBs YW5ndWFnZT0iamF2YXNjcmlwdCI+CmZ1bmN0aW9uIGluaXQoKXsKICBpZiAod2luZG93LmxheW91 dFRlc3RDb250cm9sbGVyKSB7CiAgICAgIGxheW91dFRlc3RDb250cm9sbGVyLmR1bXBBc1RleHQo KTsKICB9CiAgdmFyIGFyeTE9WzEsMl07CiAgYXJ5MS5wdXNoKGFyeTEpOwogIGRvY3VtZW50Lmdl dEVsZW1lbnRCeUlkKCJyMGMxIikuZmlyc3RDaGlsZC5ub2RlVmFsdWUgPSBhcnkxOwogIGFyeTE9 WzEsMl07CiAgdmFyIGFyeTI9WzMsNF07CiAgYXJ5MS5wdXNoKGFyeTIpOwogIGFyeTIucHVzaChh cnkxKTsKICBkb2N1bWVudC5nZXRFbGVtZW50QnlJZCgicjFjMSIpLmZpcnN0Q2hpbGQubm9kZVZh bHVlID0gYXJ5MTsKICBhcnkxLnB1c2goNSk7CiAgZG9jdW1lbnQuZ2V0RWxlbWVudEJ5SWQoInIy YzEiKS5maXJzdENoaWxkLm5vZGVWYWx1ZSA9IGFyeTE7Cn0KPC9zY3JpcHQ+CjwvaGVhZD4KPGJv ZHkgb25sb2FkPSJpbml0KCkiPgogICAgVGhpcyB3aWxsIHRlc3QgQXJyYXkudG9TdHJpbmcgd2l0 aCBjaXJjdWxhciByZWZlcmVuY2VzLiAgSWYgYW4gZWxlbWVudCBjb250YWlucyBhIHJlZmVyZW5j ZSB0byAKICAgIGl0c2VsZiBvciBvbmUgb2YgaXRzIGNoaWxkcmVuIChhdCBhbnkgZGVwdGgpIGNv bnRhaW5zIGEgcmVmZXJlbmNlIHRvIGl0LCBpdCB3aWxsIGJlIHNraXBwZWQuCiAgICBUaGlzIGNh biByZXN1bHQgaW4gZWl0aGVyIGEgdHJhaWxpbmcgJywnIChpZiB0aGUgc2VsZiByZWZlcmVuY2Ug aXMgYXQgdGhlIGVuZCBvZiB0aGUgYXJyYXkpIG9yICcsLCcgCiAgICBpZiB0aGUgc2VsZiByZWZl cmVuY2UgaXMgY29udGFpbmVkIGF0IHNvbWUgbWlkIHBvaW50IGluIHRoZSBhcnJheS4KICAgIDx0 YWJsZSBjZWxscGFkZGluZz0iMiIgY2VsbHNwYWNpbmc9IjIiIGJvcmRlcj0iMSI+CgkJPHRib2R5 PgoJCQk8dHI+CgkJCQk8dGQ+QXJyYXkgQ29udGVudHM8L3RkPjx0ZD5FeHBlY3RlZCBPdXRwdXQ8 L3RkPjx0ZD5BY3R1YWwgT3V0cHV0PC90ZD4KCQkJPC90cj4KCQkJPHRyPgoJCQkJPHRkPlsxLCAy LCBzZWxmXTwvdGQ+PHRkIGlkPSJyMGMwIj4xLDIsPC90ZD48dGQgaWQ9InIwYzEiPiZuYnNwOzwv dGQ+CgkJCTwvdHI+CgkJCTx0cj4KCQkJCTx0ZD5bMSwgMiwgWzMsIDQsIHBhcmVudF1dPC90ZD48 dGQgaWQ9InIxYzAiPjEsMiwzLDQsPC90ZD48dGQgaWQ9InIxYzEiPiZuYnNwOzwvdGQ+CgkJCTwv dHI+CgkJCTx0cj4KCQkJCTx0ZD5bMSwgMiwgWzMsIDQsIHBhcmVudF0sIDVdPC90ZD48dGQgaWQ9 InIyYzAiPjEsMiwzLDQsLDU8L3RkPjx0ZCBpZD0icjJjMSI+Jm5ic3A7PC90ZD4KCQkJPC90cj4K CQk8L3Rib2R5PgoJPC90YWJsZT4KPGRpdiBpZD0nY29uc29sZScvPgo8L2JvZHk+CjwvaHRtbD4= 4864 2005-11-29 18:29:29 -0800 2005-11-29 18:29:29 -0800 Expected output from testcase cyclicreftest-expected.txt text/plain 512 oliver VGhpcyB3aWxsIHRlc3QgQXJyYXkudG9TdHJpbmcgd2l0aCBjaXJjdWxhciByZWZlcmVuY2VzLiBJ ZiBhbiBlbGVtZW50IGNvbnRhaW5zIGEgcmVmZXJlbmNlIHRvIGl0c2VsZiBvciBvbmUgb2YgaXRz IGNoaWxkcmVuIChhdCBhbnkgZGVwdGgpIGNvbnRhaW5zIGEgcmVmZXJlbmNlIHRvIGl0LCBpdCB3 aWxsIGJlIHNraXBwZWQuIFRoaXMgY2FuIHJlc3VsdCBpbiBlaXRoZXIgYSB0cmFpbGluZyAnLCcg KGlmIHRoZSBzZWxmIHJlZmVyZW5jZSBpcyBhdCB0aGUgZW5kIG9mIHRoZSBhcnJheSkgb3IgJyws JyBpZiB0aGUgc2VsZiByZWZlcmVuY2UgaXMgY29udGFpbmVkIGF0IHNvbWUgbWlkIHBvaW50IGlu IHRoZSBhcnJheS4KQXJyYXkgQ29udGVudHMJRXhwZWN0ZWQgT3V0cHV0CUFjdHVhbCBPdXRwdXQK WzEsIDIsIHNlbGZdCTEsMiwJMSwyLApbMSwgMiwgWzMsIDQsIHBhcmVudF1dCTEsMiwzLDQsCTEs MiwzLDQsClsxLCAyLCBbMywgNCwgcGFyZW50XSwgNV0JMSwyLDMsNCwsNQkxLDIsMyw0LCw1Cgo=