)]}'
{"specs/newton/approved/dependency-driven-resource-deallocation.rst":[{"author":{"_account_id":7226,"name":"Stan Lagun","email":"stan.lagun@gmail.com","username":"istalker2"},"change_message_id":"ce625651b087d147478a739eb47c5fc290025957","unresolved":false,"context_lines":[{"line_number":26,"context_line":"methods called ``.destroy()`` which may be defined in each MuranoPL class and"},{"line_number":27,"context_line":"are intended to contain the custom code to deallocate resources allocated by"},{"line_number":28,"context_line":"the objects of those classes."},{"line_number":29,"context_line":"These methods get executed when their objects leave object graph, however the"},{"line_number":30,"context_line":"exact order of these executions is currently undefined."},{"line_number":31,"context_line":""},{"line_number":32,"context_line":"There are two different scenarios when the objects may leave the object graph"}],"source_content_type":"text/x-rst","patch_set":4,"id":"1aa78d24_2283ea99","line":29,"updated":"2016-07-05 18:30:46.000000000","message":"We use the term Object Model everywhere else for this. Though object graph is also a correct name for it inconsistent terminology makes it harder to understand","commit_id":"6a723208b446fea825f5433c2a4559eef560442a"},{"author":{"_account_id":8127,"name":"Alexander Tivelkov","email":"locher@yandex.ru","username":"ativelkov"},"change_message_id":"23fca0553146cbe6977cf9b59bb852eb51b236e5","unresolved":false,"context_lines":[{"line_number":26,"context_line":"methods called ``.destroy()`` which may be defined in each MuranoPL class and"},{"line_number":27,"context_line":"are intended to contain the custom code to deallocate resources allocated by"},{"line_number":28,"context_line":"the objects of those classes."},{"line_number":29,"context_line":"These methods get executed when their objects leave object graph, however the"},{"line_number":30,"context_line":"exact order of these executions is currently undefined."},{"line_number":31,"context_line":""},{"line_number":32,"context_line":"There are two different scenarios when the objects may leave the object graph"}],"source_content_type":"text/x-rst","patch_set":4,"id":"1aa78d24_6e24b9ae","line":29,"in_reply_to":"1aa78d24_2283ea99","updated":"2016-07-05 19:22:10.000000000","message":"Still object graph is more precise, especially in the context of GC which is all about graph-represetation of the heap","commit_id":"6a723208b446fea825f5433c2a4559eef560442a"},{"author":{"_account_id":7226,"name":"Stan Lagun","email":"stan.lagun@gmail.com","username":"istalker2"},"change_message_id":"ce625651b087d147478a739eb47c5fc290025957","unresolved":false,"context_lines":[{"line_number":53,"context_line":"   the store but is not present in the output version of the object graph is"},{"line_number":54,"context_line":"   considered to be \"orphan\" and thus its ``.destroy`` method is called. The"},{"line_number":55,"context_line":"   order of these calls is currently undefined: the objects get destroyed based"},{"line_number":56,"context_line":"   on their position in the object store, which is hardly predictable."},{"line_number":57,"context_line":""},{"line_number":58,"context_line":"Such a design is insufficient for production grade applications which"},{"line_number":59,"context_line":"often require the following scenarios:"}],"source_content_type":"text/x-rst","patch_set":4,"id":"1aa78d24_e84a61ab","line":56,"updated":"2016-07-05 18:30:46.000000000","message":"Will it be possible to destroy the object explicitly (destroy($obj))?","commit_id":"6a723208b446fea825f5433c2a4559eef560442a"},{"author":{"_account_id":8127,"name":"Alexander Tivelkov","email":"locher@yandex.ru","username":"ativelkov"},"change_message_id":"23fca0553146cbe6977cf9b59bb852eb51b236e5","unresolved":false,"context_lines":[{"line_number":53,"context_line":"   the store but is not present in the output version of the object graph is"},{"line_number":54,"context_line":"   considered to be \"orphan\" and thus its ``.destroy`` method is called. The"},{"line_number":55,"context_line":"   order of these calls is currently undefined: the objects get destroyed based"},{"line_number":56,"context_line":"   on their position in the object store, which is hardly predictable."},{"line_number":57,"context_line":""},{"line_number":58,"context_line":"Such a design is insufficient for production grade applications which"},{"line_number":59,"context_line":"often require the following scenarios:"}],"source_content_type":"text/x-rst","patch_set":4,"id":"1aa78d24_4e9c5dc0","line":56,"in_reply_to":"1aa78d24_e84a61ab","updated":"2016-07-05 19:22:10.000000000","message":"Well, this would be a GC.destroy($obj) then. But it will raise questions. For example, what should happen with the references to this object?","commit_id":"6a723208b446fea825f5433c2a4559eef560442a"},{"author":{"_account_id":7226,"name":"Stan Lagun","email":"stan.lagun@gmail.com","username":"istalker2"},"change_message_id":"ce625651b087d147478a739eb47c5fc290025957","unresolved":false,"context_lines":[{"line_number":105,"context_line":""},{"line_number":106,"context_line":"  .. note::"},{"line_number":107,"context_line":""},{"line_number":108,"context_line":"    Destruction dependencies should be persisted between deployments, however"},{"line_number":109,"context_line":"    they should be treated as \"weak links\": they obviously should not prevent"},{"line_number":110,"context_line":"    the objects from being garbage collected if they have no regular"},{"line_number":111,"context_line":"    external references."}],"source_content_type":"text/x-rst","patch_set":4,"id":"1aa78d24_edba7309","line":108,"updated":"2016-07-05 18:30:46.000000000","message":"Why should they be persisted rather than reestablished each time (in .init method, for example)?","commit_id":"6a723208b446fea825f5433c2a4559eef560442a"},{"author":{"_account_id":8127,"name":"Alexander Tivelkov","email":"locher@yandex.ru","username":"ativelkov"},"change_message_id":"23fca0553146cbe6977cf9b59bb852eb51b236e5","unresolved":false,"context_lines":[{"line_number":105,"context_line":""},{"line_number":106,"context_line":"  .. note::"},{"line_number":107,"context_line":""},{"line_number":108,"context_line":"    Destruction dependencies should be persisted between deployments, however"},{"line_number":109,"context_line":"    they should be treated as \"weak links\": they obviously should not prevent"},{"line_number":110,"context_line":"    the objects from being garbage collected if they have no regular"},{"line_number":111,"context_line":"    external references."}],"source_content_type":"text/x-rst","patch_set":4,"id":"1aa78d24_8e100503","line":108,"in_reply_to":"1aa78d24_edba7309","updated":"2016-07-05 19:22:10.000000000","message":"Imagine there is some factory which produces objects and has an action which has to be done prior to the destruction of these objects. Produced objects are persisted in the graph, same is the factory. \nIf the factory needs to re-establish the dependencies on each .init then it has to remember the set of generated objects, either as output properties or as attributes. The first has side-effects on garbage collection itself, the second is actually the same thing as the one proposed here but requiring more actions from the developer.","commit_id":"6a723208b446fea825f5433c2a4559eef560442a"},{"author":{"_account_id":7226,"name":"Stan Lagun","email":"stan.lagun@gmail.com","username":"istalker2"},"change_message_id":"ce625651b087d147478a739eb47c5fc290025957","unresolved":false,"context_lines":[{"line_number":146,"context_line":"   `owns` means that the left operand object owns (probably transitively) the"},{"line_number":147,"context_line":"   right operand object and `timestamp` is a millisecond-precision time of"},{"line_number":148,"context_line":"   objects instantiation."},{"line_number":149,"context_line":""},{"line_number":150,"context_line":"3. For each object in the list notify all the objects having a destruction"},{"line_number":151,"context_line":"   dependency on it that the target object **may** be destroyed. The notified"},{"line_number":152,"context_line":"   objects may opt to establish a reference on the target object, thus"}],"source_content_type":"text/x-rst","patch_set":4,"id":"1aa78d24_6d020371","line":149,"updated":"2016-07-05 18:30:46.000000000","message":"Currently we don\u0027t track object timestamps. If you believe that we should it deserves separate section or even a spec because it is not clear who and where should set it for the objects created in UI, will object editing change the timestamp or not etc. Also this spec doesn\u0027t explain why it is important to garbage collection and why should it be taken into consideration","commit_id":"6a723208b446fea825f5433c2a4559eef560442a"},{"author":{"_account_id":8127,"name":"Alexander Tivelkov","email":"locher@yandex.ru","username":"ativelkov"},"change_message_id":"23fca0553146cbe6977cf9b59bb852eb51b236e5","unresolved":false,"context_lines":[{"line_number":146,"context_line":"   `owns` means that the left operand object owns (probably transitively) the"},{"line_number":147,"context_line":"   right operand object and `timestamp` is a millisecond-precision time of"},{"line_number":148,"context_line":"   objects instantiation."},{"line_number":149,"context_line":""},{"line_number":150,"context_line":"3. For each object in the list notify all the objects having a destruction"},{"line_number":151,"context_line":"   dependency on it that the target object **may** be destroyed. The notified"},{"line_number":152,"context_line":"   objects may opt to establish a reference on the target object, thus"}],"source_content_type":"text/x-rst","patch_set":4,"id":"1aa78d24_6ee159fe","line":149,"in_reply_to":"1aa78d24_6d020371","updated":"2016-07-05 19:22:10.000000000","message":"Well, this may be not timestamps but ids instead. I only want the \"last resort\" to be repeatable and expectable, not completely random as the current implementation of orphan collection does.","commit_id":"6a723208b446fea825f5433c2a4559eef560442a"},{"author":{"_account_id":7226,"name":"Stan Lagun","email":"stan.lagun@gmail.com","username":"istalker2"},"change_message_id":"ce625651b087d147478a739eb47c5fc290025957","unresolved":false,"context_lines":[{"line_number":152,"context_line":"   objects may opt to establish a reference on the target object, thus"},{"line_number":153,"context_line":"   preventing its destruction. Also the handlers of these notifications may"},{"line_number":154,"context_line":"   establish new destruction dependencies or remove the existing ones; this"},{"line_number":155,"context_line":"   will not be effective though until the algorithm moves to the next step."},{"line_number":156,"context_line":""},{"line_number":157,"context_line":"4. Repeat steps 1-2 to refresh the destruction list based on the possible"},{"line_number":158,"context_line":"   changes in object references and destruction dependencies occurred during"}],"source_content_type":"text/x-rst","patch_set":4,"id":"1aa78d24_88de3d3d","line":155,"updated":"2016-07-05 18:30:46.000000000","message":"Can you provide some of use cases for this steps? I.e. what useful can I do in this step that I cannot do on step 5","commit_id":"6a723208b446fea825f5433c2a4559eef560442a"},{"author":{"_account_id":8127,"name":"Alexander Tivelkov","email":"locher@yandex.ru","username":"ativelkov"},"change_message_id":"23fca0553146cbe6977cf9b59bb852eb51b236e5","unresolved":false,"context_lines":[{"line_number":152,"context_line":"   objects may opt to establish a reference on the target object, thus"},{"line_number":153,"context_line":"   preventing its destruction. Also the handlers of these notifications may"},{"line_number":154,"context_line":"   establish new destruction dependencies or remove the existing ones; this"},{"line_number":155,"context_line":"   will not be effective though until the algorithm moves to the next step."},{"line_number":156,"context_line":""},{"line_number":157,"context_line":"4. Repeat steps 1-2 to refresh the destruction list based on the possible"},{"line_number":158,"context_line":"   changes in object references and destruction dependencies occurred during"}],"source_content_type":"text/x-rst","patch_set":4,"id":"1aa78d24_6eaf9994","line":155,"in_reply_to":"1aa78d24_88de3d3d","updated":"2016-07-05 19:22:10.000000000","message":"The main example is already there: I may re-establish a reference to an object thus preventing it from being processed in step 5. The notification at 5.1 happens when the object\u0027s deletion is inevitable.","commit_id":"6a723208b446fea825f5433c2a4559eef560442a"},{"author":{"_account_id":7226,"name":"Stan Lagun","email":"stan.lagun@gmail.com","username":"istalker2"},"change_message_id":"ce625651b087d147478a739eb47c5fc290025957","unresolved":false,"context_lines":[{"line_number":212,"context_line":"        tricky because different destruction handlers (see above) may initiate"},{"line_number":213,"context_line":"        i/o operations which may cause context-switching between the"},{"line_number":214,"context_line":"        green-threads. Appropriate implementation should not allow this"},{"line_number":215,"context_line":"        (possibly by running the GC is separate physical thread)."},{"line_number":216,"context_line":""},{"line_number":217,"context_line":""},{"line_number":218,"context_line":"Code changes"}],"source_content_type":"text/x-rst","patch_set":4,"id":"1aa78d24_c827a510","line":215,"updated":"2016-07-05 18:30:46.000000000","message":"Why is it important? The only case that I see is \"new(Resource).doSomething()\" statement in case if GC can happen after the new() but before doSomething() (which will make the result of new() indestructible by placing a reference to it in $this). But this is impossible because of the fact that green-thread switching cannot happen in the middle of expression. Can you demonstrate any case when there is a real need to do anything with threads?","commit_id":"6a723208b446fea825f5433c2a4559eef560442a"},{"author":{"_account_id":8127,"name":"Alexander Tivelkov","email":"locher@yandex.ru","username":"ativelkov"},"change_message_id":"23fca0553146cbe6977cf9b59bb852eb51b236e5","unresolved":false,"context_lines":[{"line_number":212,"context_line":"        tricky because different destruction handlers (see above) may initiate"},{"line_number":213,"context_line":"        i/o operations which may cause context-switching between the"},{"line_number":214,"context_line":"        green-threads. Appropriate implementation should not allow this"},{"line_number":215,"context_line":"        (possibly by running the GC is separate physical thread)."},{"line_number":216,"context_line":""},{"line_number":217,"context_line":""},{"line_number":218,"context_line":"Code changes"}],"source_content_type":"text/x-rst","patch_set":4,"id":"1aa78d24_a23d15bd","line":215,"in_reply_to":"1aa78d24_c827a510","updated":"2016-07-05 19:22:10.000000000","message":"Ok, here is an example. Consider objects \"Foo\" and \"Bar\", Foo has a destruction dependency on Bar. \nThere are two green-threads: gt1 and gt2. \nFoo is unreferenced, Bar is referenced by local variable $bar in gt2\u0027s call stack.\ngt1 is active and calls GC.collect().  The next instruction which is going to be executed in gt2 is \"$bar: null\".\n\nGC starts, detects that Bar is alive and Foo is doomed. In its own notification handlers (advance warning, p.3) Foo may execute some IO thus causing the threads to switch. This will result $bar to be nullified and thus Bar will become unreferenced. On the next step of the GC flow it will be GCed, but Foo was not notified in advance.","commit_id":"6a723208b446fea825f5433c2a4559eef560442a"},{"author":{"_account_id":7226,"name":"Stan Lagun","email":"stan.lagun@gmail.com","username":"istalker2"},"change_message_id":"ce625651b087d147478a739eb47c5fc290025957","unresolved":false,"context_lines":[{"line_number":226,"context_line":"  above). Halts the whole deployment (including concurrently running green"},{"line_number":227,"context_line":"  threads) until its execution is completed."},{"line_number":228,"context_line":""},{"line_number":229,"context_line":"* ``dependOnDestruction(target, warn, preDestroy, postDestroy)`` - establishes"},{"line_number":230,"context_line":"  a destruction dependency from the caller to the object passed as ``target``."},{"line_number":231,"context_line":"  Method may be called several times, in this case only a single destruction"},{"line_number":232,"context_line":"  dependency will be established, however the same amount of calls to"}],"source_content_type":"text/x-rst","patch_set":4,"id":"1aa78d24_c80de53f","line":229,"updated":"2016-07-05 18:30:46.000000000","message":"Can we use shorter names for this? For example \"watchDestruction\", \"onDestruction\" etc.","commit_id":"6a723208b446fea825f5433c2a4559eef560442a"},{"author":{"_account_id":8127,"name":"Alexander Tivelkov","email":"locher@yandex.ru","username":"ativelkov"},"change_message_id":"23fca0553146cbe6977cf9b59bb852eb51b236e5","unresolved":false,"context_lines":[{"line_number":226,"context_line":"  above). Halts the whole deployment (including concurrently running green"},{"line_number":227,"context_line":"  threads) until its execution is completed."},{"line_number":228,"context_line":""},{"line_number":229,"context_line":"* ``dependOnDestruction(target, warn, preDestroy, postDestroy)`` - establishes"},{"line_number":230,"context_line":"  a destruction dependency from the caller to the object passed as ``target``."},{"line_number":231,"context_line":"  Method may be called several times, in this case only a single destruction"},{"line_number":232,"context_line":"  dependency will be established, however the same amount of calls to"}],"source_content_type":"text/x-rst","patch_set":4,"id":"1aa78d24_e25efd10","line":229,"in_reply_to":"1aa78d24_c80de53f","updated":"2016-07-05 19:22:10.000000000","message":"I am ok with \"watchDestruction\", but I don\u0027t like \"onDestruction\", since it looks like an event handler while it actually is a subscriber to an event.\nMay be we should use \"watch\" and \"forget\" instead?","commit_id":"6a723208b446fea825f5433c2a4559eef560442a"},{"author":{"_account_id":7226,"name":"Stan Lagun","email":"stan.lagun@gmail.com","username":"istalker2"},"change_message_id":"ce625651b087d147478a739eb47c5fc290025957","unresolved":false,"context_lines":[{"line_number":243,"context_line":""},{"line_number":244,"context_line":"  * ``object`` - a target object which may be/will/was destroyed."},{"line_number":245,"context_line":""},{"line_number":246,"context_line":"* ``dropDestructionDependency(target)`` - removes the destruction dependency"},{"line_number":247,"context_line":"  from the caller to  to the object passed as ``target``. Method may be called"},{"line_number":248,"context_line":"  several times without any side-effects. If ``dependOnDestruction`` was called"},{"line_number":249,"context_line":"  more than once the same (or more) amount of calls to"}],"source_content_type":"text/x-rst","patch_set":4,"id":"1aa78d24_08088d2b","line":246,"updated":"2016-07-05 18:30:46.000000000","message":"ditto","commit_id":"6a723208b446fea825f5433c2a4559eef560442a"},{"author":{"_account_id":7226,"name":"Stan Lagun","email":"stan.lagun@gmail.com","username":"istalker2"},"change_message_id":"ce625651b087d147478a739eb47c5fc290025957","unresolved":false,"context_lines":[{"line_number":263,"context_line":""},{"line_number":264,"context_line":"* ``type`` - returns the type of GC session in accordance with pp 1-3 of"},{"line_number":265,"context_line":"  \"Garbage collection executions\" block above. May be one of `preDeployment`,"},{"line_number":266,"context_line":"  `postDeployment` or `runtime` accordingly."},{"line_number":267,"context_line":""},{"line_number":268,"context_line":"* ``phase`` - returns the phase of GC session in accordance with pp 1-6 of"},{"line_number":269,"context_line":"  \"Multi-step destruction\" block above. May be either `warnings` - for p. 3,"}],"source_content_type":"text/x-rst","patch_set":4,"id":"1aa78d24_93f4aef2","line":266,"updated":"2016-07-05 18:30:46.000000000","message":"what is the difference between postDeployment and runtime from application perspective? What are the reasons for applications to distinguish them and handle differently? post-deployment GC is just a corner case of runtime GC and it may be a single call in actual implementation","commit_id":"6a723208b446fea825f5433c2a4559eef560442a"},{"author":{"_account_id":8127,"name":"Alexander Tivelkov","email":"locher@yandex.ru","username":"ativelkov"},"change_message_id":"23fca0553146cbe6977cf9b59bb852eb51b236e5","unresolved":false,"context_lines":[{"line_number":263,"context_line":""},{"line_number":264,"context_line":"* ``type`` - returns the type of GC session in accordance with pp 1-3 of"},{"line_number":265,"context_line":"  \"Garbage collection executions\" block above. May be one of `preDeployment`,"},{"line_number":266,"context_line":"  `postDeployment` or `runtime` accordingly."},{"line_number":267,"context_line":""},{"line_number":268,"context_line":"* ``phase`` - returns the phase of GC session in accordance with pp 1-6 of"},{"line_number":269,"context_line":"  \"Multi-step destruction\" block above. May be either `warnings` - for p. 3,"}],"source_content_type":"text/x-rst","patch_set":4,"id":"1aa78d24_020509bb","line":266,"in_reply_to":"1aa78d24_93f4aef2","updated":"2016-07-05 19:22:10.000000000","message":"Well, the runtime may be an optimization which may be optional in some cases.. While the postdeployment is a mandatory one.\n\nBut I don\u0027t insist: this was added just in case, we may remove it if there is no immediate need in it.","commit_id":"6a723208b446fea825f5433c2a4559eef560442a"},{"author":{"_account_id":7226,"name":"Stan Lagun","email":"stan.lagun@gmail.com","username":"istalker2"},"change_message_id":"ce625651b087d147478a739eb47c5fc290025957","unresolved":false,"context_lines":[{"line_number":278,"context_line":"  this GC session (makes sense only in the `postDelete` handlers, returns False"},{"line_number":279,"context_line":"  in all other cases)"},{"line_number":280,"context_line":""},{"line_number":281,"context_line":"* ``interrupt()`` - immediately interrupts the GC session if the type of the"},{"line_number":282,"context_line":"  session is `runtime`, does nothing otherwise. The objects which were not"},{"line_number":283,"context_line":"  processed remain in the object store and so may be re-processed and destroyed"},{"line_number":284,"context_line":"  during subsequent GC runs."}],"source_content_type":"text/x-rst","patch_set":4,"id":"1aa78d24_b3d9ea85","line":281,"updated":"2016-07-05 18:30:46.000000000","message":"what are the use cases for this method?","commit_id":"6a723208b446fea825f5433c2a4559eef560442a"},{"author":{"_account_id":8127,"name":"Alexander Tivelkov","email":"locher@yandex.ru","username":"ativelkov"},"change_message_id":"23fca0553146cbe6977cf9b59bb852eb51b236e5","unresolved":false,"context_lines":[{"line_number":278,"context_line":"  this GC session (makes sense only in the `postDelete` handlers, returns False"},{"line_number":279,"context_line":"  in all other cases)"},{"line_number":280,"context_line":""},{"line_number":281,"context_line":"* ``interrupt()`` - immediately interrupts the GC session if the type of the"},{"line_number":282,"context_line":"  session is `runtime`, does nothing otherwise. The objects which were not"},{"line_number":283,"context_line":"  processed remain in the object store and so may be re-processed and destroyed"},{"line_number":284,"context_line":"  during subsequent GC runs."}],"source_content_type":"text/x-rst","patch_set":4,"id":"1aa78d24_a21cf513","line":281,"in_reply_to":"1aa78d24_b3d9ea85","updated":"2016-07-05 19:22:10.000000000","message":"Don\u0027t know, just though it may be useful. Isn\u0027t it?","commit_id":"6a723208b446fea825f5433c2a4559eef560442a"},{"author":{"_account_id":7226,"name":"Stan Lagun","email":"stan.lagun@gmail.com","username":"istalker2"},"change_message_id":"ce625651b087d147478a739eb47c5fc290025957","unresolved":false,"context_lines":[{"line_number":409,"context_line":"  collected;"},{"line_number":410,"context_line":""},{"line_number":411,"context_line":"* objects passed to python-backed methods and unreferenced in MuranoPL are NOT"},{"line_number":412,"context_line":"  garbage collected unless their proxies are unreferenced / GC\u0027ed in python;"},{"line_number":413,"context_line":""},{"line_number":414,"context_line":"* objects assigned to local variables of the current execution frame (i.e."},{"line_number":415,"context_line":"  variables of the current method and all the caller methods in call stack)"}],"source_content_type":"text/x-rst","patch_set":4,"id":"1aa78d24_33a5fab4","line":412,"updated":"2016-07-05 18:30:46.000000000","message":"what proxies are? Also all method arguments are present in the context even if those are pythonic arguments. The only problem is that we must maintain a least of active contexts (1 per thread), but it has nothing to do with if the method is pythonic or not (btw is this is covered in this spec at all?)","commit_id":"6a723208b446fea825f5433c2a4559eef560442a"},{"author":{"_account_id":8127,"name":"Alexander Tivelkov","email":"locher@yandex.ru","username":"ativelkov"},"change_message_id":"23fca0553146cbe6977cf9b59bb852eb51b236e5","unresolved":false,"context_lines":[{"line_number":409,"context_line":"  collected;"},{"line_number":410,"context_line":""},{"line_number":411,"context_line":"* objects passed to python-backed methods and unreferenced in MuranoPL are NOT"},{"line_number":412,"context_line":"  garbage collected unless their proxies are unreferenced / GC\u0027ed in python;"},{"line_number":413,"context_line":""},{"line_number":414,"context_line":"* objects assigned to local variables of the current execution frame (i.e."},{"line_number":415,"context_line":"  variables of the current method and all the caller methods in call stack)"}],"source_content_type":"text/x-rst","patch_set":4,"id":"1aa78d24_22e125d1","line":412,"in_reply_to":"1aa78d24_33a5fab4","updated":"2016-07-05 19:22:10.000000000","message":"By proxies I mean MuranoObjectInterface wrappers around the Object.\n\nAnd by this point I don\u0027t mean just arguments of pythonic method, I mean the pythonic-backed objects. \n\nConsider the following example:\n\n- $pythonicObject: new(PythonicClass, someAttribute\u003d\u003enew(SomeClass))\n- GC.collect()\n\nWe\u0027ve crated some object (of type SomeClass) and passed it as an argument to pythonic object. It could persist it into one of its attribute. In this case the object should not be GCed even if MuranoPL has no reference into it (and we cannot analyse the attributes of the pythonic object by GC algorithgm)\n\nSpeaking of different threads: yes, it is referenced on line 206: it mentions all frames of all threads - and there is a test description for it at line 426).","commit_id":"6a723208b446fea825f5433c2a4559eef560442a"},{"author":{"_account_id":20563,"name":"Omar Shykhkerimov","email":"omar.shikhkerimov@gmail.com","username":"oshykhkerimov"},"change_message_id":"30d5ec61c0e4857d441790e92f2ea6d07153d2b6","unresolved":false,"context_lines":[{"line_number":148,"context_line":"   current object graph with its snapshot or the reference detection (see"},{"line_number":149,"context_line":"   \"Offline\" and \"Online\" changes in \"Problem description\" section)"},{"line_number":150,"context_line":""},{"line_number":151,"context_line":"2. Sort the list according to the following algorithm: for any two objects A"},{"line_number":152,"context_line":"   and B in the list:"},{"line_number":153,"context_line":""},{"line_number":154,"context_line":"    .. code::"}],"source_content_type":"text/x-rst","patch_set":5,"id":"1aa78d24_e15af780","line":151,"range":{"start_line":151,"start_character":16,"end_line":151,"end_character":17},"updated":"2016-07-13 18:28:06.000000000","message":"\u0027of detected objects\u0027 maybe, otherwise not clear what the list talk is about","commit_id":"202a6957e0547ec4b0dd9151163899664d60d162"},{"author":{"_account_id":20563,"name":"Omar Shykhkerimov","email":"omar.shikhkerimov@gmail.com","username":"oshykhkerimov"},"change_message_id":"30d5ec61c0e4857d441790e92f2ea6d07153d2b6","unresolved":false,"context_lines":[{"line_number":148,"context_line":"   current object graph with its snapshot or the reference detection (see"},{"line_number":149,"context_line":"   \"Offline\" and \"Online\" changes in \"Problem description\" section)"},{"line_number":150,"context_line":""},{"line_number":151,"context_line":"2. Sort the list according to the following algorithm: for any two objects A"},{"line_number":152,"context_line":"   and B in the list:"},{"line_number":153,"context_line":""},{"line_number":154,"context_line":"    .. code::"}],"source_content_type":"text/x-rst","patch_set":5,"id":"1aa78d24_7c498a5f","line":151,"range":{"start_line":151,"start_character":44,"end_line":151,"end_character":54},"updated":"2016-07-13 18:28:06.000000000","message":"comparator","commit_id":"202a6957e0547ec4b0dd9151163899664d60d162"},{"author":{"_account_id":8127,"name":"Alexander Tivelkov","email":"locher@yandex.ru","username":"ativelkov"},"change_message_id":"80889b5638ceac1ad9db8791a4f1134cdc9b2300","unresolved":false,"context_lines":[{"line_number":148,"context_line":"   current object graph with its snapshot or the reference detection (see"},{"line_number":149,"context_line":"   \"Offline\" and \"Online\" changes in \"Problem description\" section)"},{"line_number":150,"context_line":""},{"line_number":151,"context_line":"2. Sort the list according to the following algorithm: for any two objects A"},{"line_number":152,"context_line":"   and B in the list:"},{"line_number":153,"context_line":""},{"line_number":154,"context_line":"    .. code::"}],"source_content_type":"text/x-rst","patch_set":5,"id":"1aa78d24_c26b4b1e","line":151,"range":{"start_line":151,"start_character":44,"end_line":151,"end_character":54},"in_reply_to":"1aa78d24_7c498a5f","updated":"2016-07-14 13:32:12.000000000","message":"Done","commit_id":"202a6957e0547ec4b0dd9151163899664d60d162"},{"author":{"_account_id":8127,"name":"Alexander Tivelkov","email":"locher@yandex.ru","username":"ativelkov"},"change_message_id":"80889b5638ceac1ad9db8791a4f1134cdc9b2300","unresolved":false,"context_lines":[{"line_number":148,"context_line":"   current object graph with its snapshot or the reference detection (see"},{"line_number":149,"context_line":"   \"Offline\" and \"Online\" changes in \"Problem description\" section)"},{"line_number":150,"context_line":""},{"line_number":151,"context_line":"2. Sort the list according to the following algorithm: for any two objects A"},{"line_number":152,"context_line":"   and B in the list:"},{"line_number":153,"context_line":""},{"line_number":154,"context_line":"    .. code::"}],"source_content_type":"text/x-rst","patch_set":5,"id":"1aa78d24_6286d7dc","line":151,"range":{"start_line":151,"start_character":16,"end_line":151,"end_character":17},"in_reply_to":"1aa78d24_e15af780","updated":"2016-07-14 13:32:12.000000000","message":"Done","commit_id":"202a6957e0547ec4b0dd9151163899664d60d162"},{"author":{"_account_id":20563,"name":"Omar Shykhkerimov","email":"omar.shikhkerimov@gmail.com","username":"oshykhkerimov"},"change_message_id":"30d5ec61c0e4857d441790e92f2ea6d07153d2b6","unresolved":false,"context_lines":[{"line_number":153,"context_line":""},{"line_number":154,"context_line":"    .. code::"},{"line_number":155,"context_line":""},{"line_number":156,"context_line":"     IF (A has-a-destruction-dependency-on B)"},{"line_number":157,"context_line":"         AND (NOT B has-a-destruction-dependency-on A)"},{"line_number":158,"context_line":"     THEN A\u003eB"},{"line_number":159,"context_line":"     ELSE IF (A owns B) THEN A\u003eB"},{"line_number":160,"context_line":"     ELSE A\u003d\u003dB"},{"line_number":161,"context_line":""},{"line_number":162,"context_line":"   where `has-a-destruction-dependency-on` means that the left operand object"},{"line_number":163,"context_line":"   has a destruction dependency (probably transitive) on right operand object,"}],"source_content_type":"text/x-rst","patch_set":5,"id":"1aa78d24_5c65aeac","line":160,"range":{"start_line":156,"start_character":5,"end_line":160,"end_character":14},"updated":"2016-07-13 18:28:06.000000000","message":"how sort algorithm with this comparator will handle cyclic dependency?\nWouldn\u0027t it be better to work with directed graph?","commit_id":"202a6957e0547ec4b0dd9151163899664d60d162"},{"author":{"_account_id":8127,"name":"Alexander Tivelkov","email":"locher@yandex.ru","username":"ativelkov"},"change_message_id":"80889b5638ceac1ad9db8791a4f1134cdc9b2300","unresolved":false,"context_lines":[{"line_number":153,"context_line":""},{"line_number":154,"context_line":"    .. code::"},{"line_number":155,"context_line":""},{"line_number":156,"context_line":"     IF (A has-a-destruction-dependency-on B)"},{"line_number":157,"context_line":"         AND (NOT B has-a-destruction-dependency-on A)"},{"line_number":158,"context_line":"     THEN A\u003eB"},{"line_number":159,"context_line":"     ELSE IF (A owns B) THEN A\u003eB"},{"line_number":160,"context_line":"     ELSE A\u003d\u003dB"},{"line_number":161,"context_line":""},{"line_number":162,"context_line":"   where `has-a-destruction-dependency-on` means that the left operand object"},{"line_number":163,"context_line":"   has a destruction dependency (probably transitive) on right operand object,"}],"source_content_type":"text/x-rst","patch_set":5,"id":"1aa78d24_829dc30c","line":160,"range":{"start_line":156,"start_character":5,"end_line":160,"end_character":14},"in_reply_to":"1aa78d24_5c65aeac","updated":"2016-07-14 13:32:12.000000000","message":"Cyclic dependency will mean that both (A has-a-destruction-dependency-on B) and (B has-a-destruction-dependency-on A). This will violate first condition of this IF. So, it will either fallback to the ownership-based resolution or will consider objects equal, which will mean that no order is observed.","commit_id":"202a6957e0547ec4b0dd9151163899664d60d162"},{"author":{"_account_id":20563,"name":"Omar Shykhkerimov","email":"omar.shikhkerimov@gmail.com","username":"oshykhkerimov"},"change_message_id":"30d5ec61c0e4857d441790e92f2ea6d07153d2b6","unresolved":false,"context_lines":[{"line_number":214,"context_line":"different scenarios:"},{"line_number":215,"context_line":""},{"line_number":216,"context_line":"1. *(currently existing)* Before the deployment, destroys objects which were"},{"line_number":217,"context_line":"   present in the object graph after the previous deployment was finished are"},{"line_number":218,"context_line":"   not found in the incoming object graph of a new deployment, i.e. the ones"},{"line_number":219,"context_line":"   explicitly removed using the API."},{"line_number":220,"context_line":""}],"source_content_type":"text/x-rst","patch_set":5,"id":"1aa78d24_b75f8de3","line":217,"range":{"start_line":217,"start_character":74,"end_line":217,"end_character":77},"updated":"2016-07-13 18:28:06.000000000","message":"did you mean \u0027and\u0027?","commit_id":"202a6957e0547ec4b0dd9151163899664d60d162"},{"author":{"_account_id":8127,"name":"Alexander Tivelkov","email":"locher@yandex.ru","username":"ativelkov"},"change_message_id":"80889b5638ceac1ad9db8791a4f1134cdc9b2300","unresolved":false,"context_lines":[{"line_number":214,"context_line":"different scenarios:"},{"line_number":215,"context_line":""},{"line_number":216,"context_line":"1. *(currently existing)* Before the deployment, destroys objects which were"},{"line_number":217,"context_line":"   present in the object graph after the previous deployment was finished are"},{"line_number":218,"context_line":"   not found in the incoming object graph of a new deployment, i.e. the ones"},{"line_number":219,"context_line":"   explicitly removed using the API."},{"line_number":220,"context_line":""}],"source_content_type":"text/x-rst","patch_set":5,"id":"1aa78d24_826b03f1","line":217,"range":{"start_line":217,"start_character":74,"end_line":217,"end_character":77},"in_reply_to":"1aa78d24_b75f8de3","updated":"2016-07-14 13:32:12.000000000","message":"I meant \"...but were not found\". Thanks, fixed.","commit_id":"202a6957e0547ec4b0dd9151163899664d60d162"},{"author":{"_account_id":20563,"name":"Omar Shykhkerimov","email":"omar.shikhkerimov@gmail.com","username":"oshykhkerimov"},"change_message_id":"30d5ec61c0e4857d441790e92f2ea6d07153d2b6","unresolved":false,"context_lines":[{"line_number":224,"context_line":"   persistable (In, Out, InOut) properties of the environment or its transitive"},{"line_number":225,"context_line":"   children)."},{"line_number":226,"context_line":""},{"line_number":227,"context_line":"3. *(proposed)* During the deployment, explicitly initiated from MuranoPL code."},{"line_number":228,"context_line":"   Destroys objects which are not part of the **complete** object graph, i.e."},{"line_number":229,"context_line":"   having no references to them from any properties of the environment"},{"line_number":230,"context_line":"   (including runtime and private properties) AND not being referenced by local"}],"source_content_type":"text/x-rst","patch_set":5,"id":"1aa78d24_ebc77196","line":227,"range":{"start_line":227,"start_character":0,"end_line":227,"end_character":2},"updated":"2016-07-13 18:28:06.000000000","message":"why do we need this scenario? is there a chance that objects, that must be deleted during deployment, will not be deleted after that (second scenario)?","commit_id":"202a6957e0547ec4b0dd9151163899664d60d162"},{"author":{"_account_id":8127,"name":"Alexander Tivelkov","email":"locher@yandex.ru","username":"ativelkov"},"change_message_id":"80889b5638ceac1ad9db8791a4f1134cdc9b2300","unresolved":false,"context_lines":[{"line_number":224,"context_line":"   persistable (In, Out, InOut) properties of the environment or its transitive"},{"line_number":225,"context_line":"   children)."},{"line_number":226,"context_line":""},{"line_number":227,"context_line":"3. *(proposed)* During the deployment, explicitly initiated from MuranoPL code."},{"line_number":228,"context_line":"   Destroys objects which are not part of the **complete** object graph, i.e."},{"line_number":229,"context_line":"   having no references to them from any properties of the environment"},{"line_number":230,"context_line":"   (including runtime and private properties) AND not being referenced by local"}],"source_content_type":"text/x-rst","patch_set":5,"id":"1aa78d24_42163b59","line":227,"range":{"start_line":227,"start_character":0,"end_line":227,"end_character":2},"in_reply_to":"1aa78d24_ebc77196","updated":"2016-07-14 13:32:12.000000000","message":"Nope. But there may be a need to free the resources immediately, wihout waiting for the end of the deployment.\nImagine that you want to destroy some VMs and spawn some new ones instead, in the same deployment. If the destruction will have to wait till the end of deployment it will mean that there will be a double amount of VMs. This may overexceed the quota.\n\nAlso this is needed for the test-runner-based tests.","commit_id":"202a6957e0547ec4b0dd9151163899664d60d162"},{"author":{"_account_id":20563,"name":"Omar Shykhkerimov","email":"omar.shikhkerimov@gmail.com","username":"oshykhkerimov"},"change_message_id":"30d5ec61c0e4857d441790e92f2ea6d07153d2b6","unresolved":false,"context_lines":[{"line_number":264,"context_line":"  * ``sender`` - an instance of ``GC`` class describing the current"},{"line_number":265,"context_line":"    garbage collection session."},{"line_number":266,"context_line":""},{"line_number":267,"context_line":"  * ``object`` - a target object which may is going to be destroyed. It is not"},{"line_number":268,"context_line":"    recommended to persist the reference to this object anywhere. This will not"},{"line_number":269,"context_line":"    prevent the object from being garbage collected but the object will be"},{"line_number":270,"context_line":"    moved to the \"dead\" state which is almost always bad. The option to do so"}],"source_content_type":"text/x-rst","patch_set":5,"id":"1aa78d24_ee065f1b","line":267,"range":{"start_line":267,"start_character":43,"end_line":267,"end_character":45},"updated":"2016-07-13 18:28:06.000000000","message":"be","commit_id":"202a6957e0547ec4b0dd9151163899664d60d162"},{"author":{"_account_id":8127,"name":"Alexander Tivelkov","email":"locher@yandex.ru","username":"ativelkov"},"change_message_id":"80889b5638ceac1ad9db8791a4f1134cdc9b2300","unresolved":false,"context_lines":[{"line_number":264,"context_line":"  * ``sender`` - an instance of ``GC`` class describing the current"},{"line_number":265,"context_line":"    garbage collection session."},{"line_number":266,"context_line":""},{"line_number":267,"context_line":"  * ``object`` - a target object which may is going to be destroyed. It is not"},{"line_number":268,"context_line":"    recommended to persist the reference to this object anywhere. This will not"},{"line_number":269,"context_line":"    prevent the object from being garbage collected but the object will be"},{"line_number":270,"context_line":"    moved to the \"dead\" state which is almost always bad. The option to do so"}],"source_content_type":"text/x-rst","patch_set":5,"id":"1aa78d24_9d3c02c1","line":267,"range":{"start_line":267,"start_character":43,"end_line":267,"end_character":45},"in_reply_to":"1aa78d24_ee065f1b","updated":"2016-07-14 13:32:12.000000000","message":"nope, it should be \"is going to\". But \"may\" is a typo indeed.","commit_id":"202a6957e0547ec4b0dd9151163899664d60d162"},{"author":{"_account_id":20563,"name":"Omar Shykhkerimov","email":"omar.shikhkerimov@gmail.com","username":"oshykhkerimov"},"change_message_id":"30d5ec61c0e4857d441790e92f2ea6d07153d2b6","unresolved":false,"context_lines":[{"line_number":274,"context_line":"* ``unsubscribe(target)`` - removes the destruction dependency from the caller"},{"line_number":275,"context_line":"  to the object passed as ``target``. Method may be called several times"},{"line_number":276,"context_line":"  without any side-effects. If ``subscribe`` was called more than once the same"},{"line_number":277,"context_line":"  (or more) amount of calls to ``unsubscribe`` if needed to remove the"},{"line_number":278,"context_line":"  dependency."},{"line_number":279,"context_line":""},{"line_number":280,"context_line":"An instance of ``GC`` class will be created during the garbage collection"}],"source_content_type":"text/x-rst","patch_set":5,"id":"1aa78d24_8eeb5b12","line":277,"range":{"start_line":277,"start_character":47,"end_line":277,"end_character":49},"updated":"2016-07-13 18:28:06.000000000","message":"is","commit_id":"202a6957e0547ec4b0dd9151163899664d60d162"},{"author":{"_account_id":8127,"name":"Alexander Tivelkov","email":"locher@yandex.ru","username":"ativelkov"},"change_message_id":"80889b5638ceac1ad9db8791a4f1134cdc9b2300","unresolved":false,"context_lines":[{"line_number":274,"context_line":"* ``unsubscribe(target)`` - removes the destruction dependency from the caller"},{"line_number":275,"context_line":"  to the object passed as ``target``. Method may be called several times"},{"line_number":276,"context_line":"  without any side-effects. If ``subscribe`` was called more than once the same"},{"line_number":277,"context_line":"  (or more) amount of calls to ``unsubscribe`` if needed to remove the"},{"line_number":278,"context_line":"  dependency."},{"line_number":279,"context_line":""},{"line_number":280,"context_line":"An instance of ``GC`` class will be created during the garbage collection"}],"source_content_type":"text/x-rst","patch_set":5,"id":"1aa78d24_fd49c661","line":277,"range":{"start_line":277,"start_character":47,"end_line":277,"end_character":49},"in_reply_to":"1aa78d24_8eeb5b12","updated":"2016-07-14 13:32:12.000000000","message":"Done","commit_id":"202a6957e0547ec4b0dd9151163899664d60d162"},{"author":{"_account_id":20563,"name":"Omar Shykhkerimov","email":"omar.shikhkerimov@gmail.com","username":"oshykhkerimov"},"change_message_id":"30d5ec61c0e4857d441790e92f2ea6d07153d2b6","unresolved":false,"context_lines":[{"line_number":318,"context_line":"destruction dependencies objects will be collected based on their ownership"},{"line_number":319,"context_line":"relationships, i.e. as it is done in the current implementation."},{"line_number":320,"context_line":""},{"line_number":321,"context_line":"The packages containing classes which explicitly call the the methods of ``GC``"},{"line_number":322,"context_line":"should have package format of at least 1.4 to prevent their execution on older"},{"line_number":323,"context_line":"versions of Murano which do not have this feature."},{"line_number":324,"context_line":""}],"source_content_type":"text/x-rst","patch_set":5,"id":"1aa78d24_31c8c856","line":321,"range":{"start_line":321,"start_character":58,"end_line":321,"end_character":61},"updated":"2016-07-13 18:28:06.000000000","message":"typo","commit_id":"202a6957e0547ec4b0dd9151163899664d60d162"},{"author":{"_account_id":20563,"name":"Omar Shykhkerimov","email":"omar.shikhkerimov@gmail.com","username":"oshykhkerimov"},"change_message_id":"f016fe7f21a6ac709c3f3debbdfe7ae8dbff2582","unresolved":false,"context_lines":[{"line_number":467,"context_line":""},{"line_number":468,"context_line":"* A may cancel GC in both warning and pre-destroy notification handlers;"},{"line_number":469,"context_line":""},{"line_number":470,"context_line":"* A may establish more then 1 destruction dependencies on B and still be"},{"line_number":471,"context_line":"  notified just once;"},{"line_number":472,"context_line":""},{"line_number":473,"context_line":"* A may remove the destruction dependency and not get notified on B\u0027s"}],"source_content_type":"text/x-rst","patch_set":5,"id":"1aa78d24_7ba32601","line":470,"range":{"start_line":470,"start_character":23,"end_line":470,"end_character":54},"updated":"2016-07-14 11:42:33.000000000","message":"than 1 destruction dependency","commit_id":"202a6957e0547ec4b0dd9151163899664d60d162"},{"author":{"_account_id":8127,"name":"Alexander Tivelkov","email":"locher@yandex.ru","username":"ativelkov"},"change_message_id":"80889b5638ceac1ad9db8791a4f1134cdc9b2300","unresolved":false,"context_lines":[{"line_number":467,"context_line":""},{"line_number":468,"context_line":"* A may cancel GC in both warning and pre-destroy notification handlers;"},{"line_number":469,"context_line":""},{"line_number":470,"context_line":"* A may establish more then 1 destruction dependencies on B and still be"},{"line_number":471,"context_line":"  notified just once;"},{"line_number":472,"context_line":""},{"line_number":473,"context_line":"* A may remove the destruction dependency and not get notified on B\u0027s"}],"source_content_type":"text/x-rst","patch_set":5,"id":"1aa78d24_9d6362dd","line":470,"range":{"start_line":470,"start_character":23,"end_line":470,"end_character":54},"in_reply_to":"1aa78d24_7ba32601","updated":"2016-07-14 13:32:12.000000000","message":"Done","commit_id":"202a6957e0547ec4b0dd9151163899664d60d162"},{"author":{"_account_id":20563,"name":"Omar Shykhkerimov","email":"omar.shikhkerimov@gmail.com","username":"oshykhkerimov"},"change_message_id":"f016fe7f21a6ac709c3f3debbdfe7ae8dbff2582","unresolved":false,"context_lines":[{"line_number":477,"context_line":"  (N\u003eM) then notifications are still delivered;"},{"line_number":478,"context_line":""},{"line_number":479,"context_line":"* If A established N destruction dependencies and then removed them M times,"},{"line_number":480,"context_line":"  (N\u003cM) then notifications are not delivered;"},{"line_number":481,"context_line":""},{"line_number":482,"context_line":"* B may establish a destruction dependency on itself thus subscribing to"},{"line_number":483,"context_line":"  appropriate notifications;"}],"source_content_type":"text/x-rst","patch_set":5,"id":"1aa78d24_fbe55638","line":480,"range":{"start_line":480,"start_character":3,"end_line":480,"end_character":6},"updated":"2016-07-14 11:42:33.000000000","message":"N\u003c\u003dM","commit_id":"202a6957e0547ec4b0dd9151163899664d60d162"},{"author":{"_account_id":8127,"name":"Alexander Tivelkov","email":"locher@yandex.ru","username":"ativelkov"},"change_message_id":"80889b5638ceac1ad9db8791a4f1134cdc9b2300","unresolved":false,"context_lines":[{"line_number":477,"context_line":"  (N\u003eM) then notifications are still delivered;"},{"line_number":478,"context_line":""},{"line_number":479,"context_line":"* If A established N destruction dependencies and then removed them M times,"},{"line_number":480,"context_line":"  (N\u003cM) then notifications are not delivered;"},{"line_number":481,"context_line":""},{"line_number":482,"context_line":"* B may establish a destruction dependency on itself thus subscribing to"},{"line_number":483,"context_line":"  appropriate notifications;"}],"source_content_type":"text/x-rst","patch_set":5,"id":"1aa78d24_dd5dea16","line":480,"range":{"start_line":480,"start_character":3,"end_line":480,"end_character":6},"in_reply_to":"1aa78d24_fbe55638","updated":"2016-07-14 13:32:12.000000000","message":"Done","commit_id":"202a6957e0547ec4b0dd9151163899664d60d162"},{"author":{"_account_id":20773,"name":"Valerii Kovalchuk","email":"kovvalole@gmail.com","username":"vakovalchuk"},"change_message_id":"1f2c2a27504a8f7b5175c46ab788ea5e3a246ee9","unresolved":false,"context_lines":[{"line_number":8,"context_line":"Dependency-driven multi-step resource deallocation"},{"line_number":9,"context_line":"\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d"},{"line_number":10,"context_line":""},{"line_number":11,"context_line":"https://blueprints.launchpad.net/murano/+spec/dependency-driven-resource-deallocation"},{"line_number":12,"context_line":""},{"line_number":13,"context_line":"Murano components may allocate various kinds of resources: virtual machines,"},{"line_number":14,"context_line":"networks, volumes etc. When these components get removed from the deployment"}],"source_content_type":"text/x-rst","patch_set":6,"id":"bacf61ea_ca04dfc1","line":11,"range":{"start_line":11,"start_character":0,"end_line":11,"end_character":85},"updated":"2016-07-28 15:49:55.000000000","message":"please create corresponding blueprint","commit_id":"4f892f5b53cb14a90e7f4f6057575a85cefa0db4"},{"author":{"_account_id":7226,"name":"Stan Lagun","email":"stan.lagun@gmail.com","username":"istalker2"},"change_message_id":"b79216f058d4b2a3071e61f2ddb91ddf9807b349","unresolved":false,"context_lines":[{"line_number":69,"context_line":"* Sometimes the actions being executed during the destruction of an object may"},{"line_number":70,"context_line":"  depend on the fact whether some other object is about to be deleted or not."},{"line_number":71,"context_line":""},{"line_number":72,"context_line":"* Sometimes when an object is about to be removed from the object graph because"},{"line_number":73,"context_line":"  of unreferencing some other object may choose to establish a reference to it"},{"line_number":74,"context_line":"  and thus prevent the former object from being destroyed."},{"line_number":75,"context_line":""}],"source_content_type":"text/x-rst","patch_set":6,"id":"1aa78d24_4c0dbb34","line":72,"updated":"2016-07-14 20:22:42.000000000","message":"This item doesn\u0027t seem to be a requirement for \"production grade applications\"","commit_id":"4f892f5b53cb14a90e7f4f6057575a85cefa0db4"},{"author":{"_account_id":7226,"name":"Stan Lagun","email":"stan.lagun@gmail.com","username":"istalker2"},"change_message_id":"b79216f058d4b2a3071e61f2ddb91ddf9807b349","unresolved":false,"context_lines":[{"line_number":128,"context_line":"    These should introduce a change in the format being used to serialize the"},{"line_number":129,"context_line":"    object graph into json representation: a new section to contain the"},{"line_number":130,"context_line":"    destruction dependencies should be added."},{"line_number":131,"context_line":""},{"line_number":132,"context_line":""},{"line_number":133,"context_line":"Since the destruction dependencies are directional there is a theoretical"},{"line_number":134,"context_line":"possibility of a circular dependency to exist. In case if two or more objects"}],"source_content_type":"text/x-rst","patch_set":6,"id":"1aa78d24_0cdf333b","line":131,"updated":"2016-07-14 20:22:42.000000000","message":"This requires some more explanation:\n1) Why should they be persisted (rather than reestablished each time)? Because it is not obvious to say the least I\u0027d like to see reasons behind such decision\n2) What exactly the format they get serialized to? Where are you going to put it?\n3) Where exactly those dependencies are going to be established? We don\u0027t have any method that is called upon the first object creation. \".init\" is called each time an object is loaded (which is every time you deploy the environment for objects from the object model). So if you going to establish dependencies there they are going to be reestablished each time. Then what\u0027s the point of persisting them?","commit_id":"4f892f5b53cb14a90e7f4f6057575a85cefa0db4"},{"author":{"_account_id":7226,"name":"Stan Lagun","email":"stan.lagun@gmail.com","username":"istalker2"},"change_message_id":"b79216f058d4b2a3071e61f2ddb91ddf9807b349","unresolved":false,"context_lines":[{"line_number":178,"context_line":""},{"line_number":179,"context_line":"   3.2. Call the ``.destroy()`` method of the object if it is present."},{"line_number":180,"context_line":""},{"line_number":181,"context_line":"   3.3. Change the object\u0027s status to \"Dead\" (see below)."},{"line_number":182,"context_line":""},{"line_number":183,"context_line":"Dead objects"},{"line_number":184,"context_line":"^^^^^^^^^^^^"}],"source_content_type":"text/x-rst","patch_set":6,"id":"1aa78d24_0cb293e5","line":181,"range":{"start_line":181,"start_character":39,"end_line":181,"end_character":43},"updated":"2016-07-14 20:22:42.000000000","message":"can we use some other term for it like \"reclaimed\", \"disposed\" etc. It is not dead if you can access its data.","commit_id":"4f892f5b53cb14a90e7f4f6057575a85cefa0db4"},{"author":{"_account_id":7535,"name":"Nikolay Starodubtsev","email":"starodubcevna@gmail.com","username":"Kezar"},"change_message_id":"469b4156c075a444c5e0d1a93ce688f66b8f7818","unresolved":false,"context_lines":[{"line_number":178,"context_line":""},{"line_number":179,"context_line":"   3.2. Call the ``.destroy()`` method of the object if it is present."},{"line_number":180,"context_line":""},{"line_number":181,"context_line":"   3.3. Change the object\u0027s status to \"Dead\" (see below)."},{"line_number":182,"context_line":""},{"line_number":183,"context_line":"Dead objects"},{"line_number":184,"context_line":"^^^^^^^^^^^^"}],"source_content_type":"text/x-rst","patch_set":6,"id":"dada55a8_6dbe2800","line":181,"range":{"start_line":181,"start_character":39,"end_line":181,"end_character":43},"updated":"2016-07-21 09:30:03.000000000","message":"maybe \"active/inactive\". as for me it will define status of the object better.","commit_id":"4f892f5b53cb14a90e7f4f6057575a85cefa0db4"},{"author":{"_account_id":7226,"name":"Stan Lagun","email":"stan.lagun@gmail.com","username":"istalker2"},"change_message_id":"b79216f058d4b2a3071e61f2ddb91ddf9807b349","unresolved":false,"context_lines":[{"line_number":201,"context_line":"the object will have a special flag in their class header (the \"?\" section) to"},{"line_number":202,"context_line":"indicate their special status. When deserialized from json such objects will"},{"line_number":203,"context_line":"retain their \"dead\" status, so the method execution will still be impossible"},{"line_number":204,"context_line":"even in subsequent deployments."},{"line_number":205,"context_line":""},{"line_number":206,"context_line":"When the dead objects are unreferenced from the object graph they go away"},{"line_number":207,"context_line":"without additional actions: garbage collector ignores them since their"}],"source_content_type":"text/x-rst","patch_set":6,"id":"1aa78d24_4ccbfb4a","line":204,"updated":"2016-07-14 20:22:42.000000000","message":"We discussed that it should be still possible to read property values of such object","commit_id":"4f892f5b53cb14a90e7f4f6057575a85cefa0db4"},{"author":{"_account_id":7535,"name":"Nikolay Starodubtsev","email":"starodubcevna@gmail.com","username":"Kezar"},"change_message_id":"c86248ce435f0ab0970a31b0432132202efed8f0","unresolved":false,"context_lines":[{"line_number":201,"context_line":"the object will have a special flag in their class header (the \"?\" section) to"},{"line_number":202,"context_line":"indicate their special status. When deserialized from json such objects will"},{"line_number":203,"context_line":"retain their \"dead\" status, so the method execution will still be impossible"},{"line_number":204,"context_line":"even in subsequent deployments."},{"line_number":205,"context_line":""},{"line_number":206,"context_line":"When the dead objects are unreferenced from the object graph they go away"},{"line_number":207,"context_line":"without additional actions: garbage collector ignores them since their"}],"source_content_type":"text/x-rst","patch_set":6,"id":"dada55a8_e20ee739","line":204,"in_reply_to":"1aa78d24_4ccbfb4a","updated":"2016-07-20 09:28:22.000000000","message":"I guess it\u0027s covered in this sentence:\n\n    MuranoPL executor will not allow to execute any methods on the dead objects, however their properties remain accessible (i.e. readable) so any runtime information associated with them may be recovered.\n\nAm I wrong?","commit_id":"4f892f5b53cb14a90e7f4f6057575a85cefa0db4"},{"author":{"_account_id":7535,"name":"Nikolay Starodubtsev","email":"starodubcevna@gmail.com","username":"Kezar"},"change_message_id":"c86248ce435f0ab0970a31b0432132202efed8f0","unresolved":false,"context_lines":[{"line_number":213,"context_line":"The multi-step object destruction described above should take place in three"},{"line_number":214,"context_line":"different scenarios:"},{"line_number":215,"context_line":""},{"line_number":216,"context_line":"1. *(currently existing)* Before the deployment, destroys objects which were"},{"line_number":217,"context_line":"   present in the object graph after the previous deployment was finished but"},{"line_number":218,"context_line":"   were not found in the incoming object graph of a new deployment, i.e. the"},{"line_number":219,"context_line":"   ones explicitly removed using the API."},{"line_number":220,"context_line":""},{"line_number":221,"context_line":"2. *(currently existing)* After the deployment, destroys the objects existing"},{"line_number":222,"context_line":"   in the Object Store but not being a part of the **persistable** object graph"},{"line_number":223,"context_line":"   of current environment, i.e. having no references to them from the"},{"line_number":224,"context_line":"   persistable (In, Out, InOut) properties of the environment or its transitive"},{"line_number":225,"context_line":"   children)."},{"line_number":226,"context_line":""},{"line_number":227,"context_line":"3. *(proposed)* During the deployment, explicitly initiated from MuranoPL code."},{"line_number":228,"context_line":"   Destroys objects which are not part of the **complete** object graph, i.e."}],"source_content_type":"text/x-rst","patch_set":6,"id":"dada55a8_b8dc2074","line":225,"range":{"start_line":216,"start_character":0,"end_line":225,"end_character":13},"updated":"2016-07-20 09:28:22.000000000","message":"Am I right and we need to \"attach\" existing scenarios to GC process?","commit_id":"4f892f5b53cb14a90e7f4f6057575a85cefa0db4"},{"author":{"_account_id":7226,"name":"Stan Lagun","email":"stan.lagun@gmail.com","username":"istalker2"},"change_message_id":"b79216f058d4b2a3071e61f2ddb91ddf9807b349","unresolved":false,"context_lines":[{"line_number":225,"context_line":"   children)."},{"line_number":226,"context_line":""},{"line_number":227,"context_line":"3. *(proposed)* During the deployment, explicitly initiated from MuranoPL code."},{"line_number":228,"context_line":"   Destroys objects which are not part of the **complete** object graph, i.e."},{"line_number":229,"context_line":"   having no references to them from any properties of the environment"},{"line_number":230,"context_line":"   (including runtime and private properties) AND not being referenced by local"},{"line_number":231,"context_line":"   variables in any frame of all the the green threads of current deployment."}],"source_content_type":"text/x-rst","patch_set":6,"id":"1aa78d24_0c8573ce","line":228,"updated":"2016-07-14 20:22:42.000000000","message":"I think that the algorithm to obtain such complete graph should also be in this spec. Especially considering that it is not obvious with all those multi-thread, multi-context issues","commit_id":"4f892f5b53cb14a90e7f4f6057575a85cefa0db4"},{"author":{"_account_id":7535,"name":"Nikolay Starodubtsev","email":"starodubcevna@gmail.com","username":"Kezar"},"change_message_id":"c86248ce435f0ab0970a31b0432132202efed8f0","unresolved":false,"context_lines":[{"line_number":225,"context_line":"   children)."},{"line_number":226,"context_line":""},{"line_number":227,"context_line":"3. *(proposed)* During the deployment, explicitly initiated from MuranoPL code."},{"line_number":228,"context_line":"   Destroys objects which are not part of the **complete** object graph, i.e."},{"line_number":229,"context_line":"   having no references to them from any properties of the environment"},{"line_number":230,"context_line":"   (including runtime and private properties) AND not being referenced by local"},{"line_number":231,"context_line":"   variables in any frame of all the the green threads of current deployment."}],"source_content_type":"text/x-rst","patch_set":6,"id":"dada55a8_78b748f5","line":228,"in_reply_to":"1aa78d24_0c8573ce","updated":"2016-07-20 09:28:22.000000000","message":"Partially agree here.","commit_id":"4f892f5b53cb14a90e7f4f6057575a85cefa0db4"},{"author":{"_account_id":7535,"name":"Nikolay Starodubtsev","email":"starodubcevna@gmail.com","username":"Kezar"},"change_message_id":"c86248ce435f0ab0970a31b0432132202efed8f0","unresolved":false,"context_lines":[{"line_number":354,"context_line":"  ativelkov"},{"line_number":355,"context_line":""},{"line_number":356,"context_line":"Other contributors:"},{"line_number":357,"context_line":"  TBD"},{"line_number":358,"context_line":""},{"line_number":359,"context_line":"Work Items"},{"line_number":360,"context_line":"----------"}],"source_content_type":"text/x-rst","patch_set":6,"id":"dada55a8_c5e64d14","line":357,"range":{"start_line":357,"start_character":2,"end_line":357,"end_character":5},"updated":"2016-07-20 09:28:22.000000000","message":"I want to help with things described in the spec. Please, mark me here \u003d)","commit_id":"4f892f5b53cb14a90e7f4f6057575a85cefa0db4"},{"author":{"_account_id":7535,"name":"Nikolay Starodubtsev","email":"starodubcevna@gmail.com","username":"Kezar"},"change_message_id":"b6bec54f9a433cdaf82ed2eb2e0d3fe274ba82a4","unresolved":false,"context_lines":[{"line_number":364,"context_line":"* Introduce changes to MuranoObject class to keep track of \"dead or alive\""},{"line_number":365,"context_line":"  object status."},{"line_number":366,"context_line":""},{"line_number":367,"context_line":"* Modify the serializer / deserializer to properly persist the value of the"},{"line_number":368,"context_line":"  \"dead object\" flag as well as persistence of destruction dependencies of"},{"line_number":369,"context_line":"  objects."},{"line_number":370,"context_line":""},{"line_number":371,"context_line":"* Modify the code which instantiates yaql contexts for MuranoPL so all the"},{"line_number":372,"context_line":"  created contexts are tracked by the execution session."}],"source_content_type":"text/x-rst","patch_set":6,"id":"dada55a8_96ca9219","line":369,"range":{"start_line":367,"start_character":0,"end_line":369,"end_character":10},"updated":"2016-07-21 09:10:43.000000000","message":"I just realize that this step can be split into two and joined to first two steps. What I mean:\n\n    * Implement a system to define and persist destruction dependencies.Modify the serializer / deserializer to handle destruction dependencies.\n    * Introduce changes to MuranoObject class to keep track of \"dead or alive\" object status. Modify the serializer / deserializer to properly persist the value of the \"dead object\" flag.\n\n\nWhat do you think about it? It\u0027s just hard for me to realize how to prove that first two works if we do not modify serializer. Correct me if I wrong.","commit_id":"4f892f5b53cb14a90e7f4f6057575a85cefa0db4"},{"author":{"_account_id":7535,"name":"Nikolay Starodubtsev","email":"starodubcevna@gmail.com","username":"Kezar"},"change_message_id":"1be47a3adf7430208011726bcb78c64e54668811","unresolved":false,"context_lines":[{"line_number":374,"context_line":"* Implement sorting algorithms to arrange objects-to-be-destroyed based on"},{"line_number":375,"context_line":"  criteria defined in p.2 of \"Multi-step destruction\" section above."},{"line_number":376,"context_line":""},{"line_number":377,"context_line":"* Modify an algorithm to collect alive object roots from the runtime and"},{"line_number":378,"context_line":"  private properties and local variables of executing threads."},{"line_number":379,"context_line":""},{"line_number":380,"context_line":"* Implement multi-step destruction workflow."}],"source_content_type":"text/x-rst","patch_set":6,"id":"dada55a8_55f8cfc4","line":377,"range":{"start_line":377,"start_character":33,"end_line":377,"end_character":38},"updated":"2016-07-22 09:46:01.000000000","message":"may be I misunderstood this item, but why we want to collect alive roots, instead of collecting unreachable objects as it done in Python, AFAIK.","commit_id":"4f892f5b53cb14a90e7f4f6057575a85cefa0db4"},{"author":{"_account_id":7535,"name":"Nikolay Starodubtsev","email":"starodubcevna@gmail.com","username":"Kezar"},"change_message_id":"1be47a3adf7430208011726bcb78c64e54668811","unresolved":false,"context_lines":[{"line_number":416,"context_line":"  collected;"},{"line_number":417,"context_line":""},{"line_number":418,"context_line":"* objects passed to python-backed objects and unreferenced in MuranoPL are NOT"},{"line_number":419,"context_line":"  garbage collected unless their MoranoObjectInterface proxies are unreferenced"},{"line_number":420,"context_line":"  / GC\u0027ed in python;"},{"line_number":421,"context_line":""},{"line_number":422,"context_line":"* objects assigned to local variables of the current execution frame (i.e."}],"source_content_type":"text/x-rst","patch_set":6,"id":"dada55a8_525d8547","line":419,"range":{"start_line":419,"start_character":33,"end_line":419,"end_character":54},"updated":"2016-07-22 09:46:01.000000000","message":"typo. Probably MuranoObjectInterface","commit_id":"4f892f5b53cb14a90e7f4f6057575a85cefa0db4"},{"author":{"_account_id":20773,"name":"Valerii Kovalchuk","email":"kovvalole@gmail.com","username":"vakovalchuk"},"change_message_id":"8b46df94cb469a54d6a25f78eef6e4d66404f85b","unresolved":false,"context_lines":[{"line_number":126,"context_line":"developed as part of the \"Application Development Framework\" initiative [1] is"},{"line_number":127,"context_line":"a collection of ``Instance`` objects which are dynamically created in runtime."},{"line_number":128,"context_line":"These instances are persisted in a `InOut`-typed property of this grouping"},{"line_number":129,"context_line":"class. In various circumstances this class mat need to establish destruction"},{"line_number":130,"context_line":"dependencies on its member-instances. It\u0027s convenient to establish such"},{"line_number":131,"context_line":"dependencies right at the moment when the object is created. If the"},{"line_number":132,"context_line":"dependencies are not persisted, they would have to be recreated on every"}],"source_content_type":"text/x-rst","patch_set":9,"id":"bacf61ea_14cfd556","line":129,"range":{"start_line":129,"start_character":43,"end_line":129,"end_character":46},"updated":"2016-08-05 08:15:20.000000000","message":"may","commit_id":"38b114906a41199a8b7655fe6f2f01b6f149f9a8"},{"author":{"_account_id":20773,"name":"Valerii Kovalchuk","email":"kovvalole@gmail.com","username":"vakovalchuk"},"change_message_id":"8b46df94cb469a54d6a25f78eef6e4d66404f85b","unresolved":false,"context_lines":[{"line_number":260,"context_line":""},{"line_number":261,"context_line":"To implement p.3 above a new algorithm is needed. It should analyze all the"},{"line_number":262,"context_line":"active contexts of all the running greenthreads of the current deployment and"},{"line_number":263,"context_line":"retrieve all the regular data variables from that context (i.e. the ones with"},{"line_number":264,"context_line":"regular alphanumeric names, skipping the ones starting with ? etc), traversing"},{"line_number":265,"context_line":"through the parent contexts as well. All the objects of ``MuranoObject`` type"},{"line_number":266,"context_line":"collected this way should be added to the \"queue of active roots\" to be used"},{"line_number":267,"context_line":"for further processing."}],"source_content_type":"text/x-rst","patch_set":9,"id":"bacf61ea_eff4eef1","line":264,"range":{"start_line":263,"start_character":59,"end_line":264,"end_character":65},"updated":"2016-08-05 08:15:20.000000000","message":"do we leave that requirement?","commit_id":"38b114906a41199a8b7655fe6f2f01b6f149f9a8"},{"author":{"_account_id":20773,"name":"Valerii Kovalchuk","email":"kovvalole@gmail.com","username":"vakovalchuk"},"change_message_id":"8b46df94cb469a54d6a25f78eef6e4d66404f85b","unresolved":false,"context_lines":[{"line_number":277,"context_line":"considered as candidates for garbage collection."},{"line_number":278,"context_line":""},{"line_number":279,"context_line":"There should be one additional check made before doing the actual destruction:"},{"line_number":280,"context_line":"some of them objects may have no valid access paths from MuranoPL objects, but"},{"line_number":281,"context_line":"could be referenced from some python-back objects. This happens when an Object"},{"line_number":282,"context_line":"is passed to some method backed with pythonic code. In this case the executor"},{"line_number":283,"context_line":"creates an object of type ``MuranoObjectInterface`` - a wrapper to simplify the"}],"source_content_type":"text/x-rst","patch_set":9,"id":"bacf61ea_4f21ba4c","line":280,"range":{"start_line":280,"start_character":8,"end_line":280,"end_character":12},"updated":"2016-08-05 08:15:20.000000000","message":"the","commit_id":"38b114906a41199a8b7655fe6f2f01b6f149f9a8"},{"author":{"_account_id":20773,"name":"Valerii Kovalchuk","email":"kovvalole@gmail.com","username":"vakovalchuk"},"change_message_id":"8b46df94cb469a54d6a25f78eef6e4d66404f85b","unresolved":false,"context_lines":[{"line_number":353,"context_line":""},{"line_number":354,"context_line":"    .. note::"},{"line_number":355,"context_line":""},{"line_number":356,"context_line":"        The is the opposite of how the destruction dependencies are stored when"},{"line_number":357,"context_line":"        the model is serialized: in serialized form that is the dependent"},{"line_number":358,"context_line":"        object who owns the reference to the dependency object. In runtime it"},{"line_number":359,"context_line":"        is the dependency  object who owns the reference to dependent object."}],"source_content_type":"text/x-rst","patch_set":9,"id":"bacf61ea_8a5280dc","line":356,"range":{"start_line":356,"start_character":8,"end_line":356,"end_character":11},"updated":"2016-08-05 08:15:20.000000000","message":"This","commit_id":"38b114906a41199a8b7655fe6f2f01b6f149f9a8"},{"author":{"_account_id":7226,"name":"Stan Lagun","email":"stan.lagun@gmail.com","username":"istalker2"},"change_message_id":"15a69db088ba5fe807f2399d70c28ee760fb1267","unresolved":false,"context_lines":[{"line_number":143,"context_line":"    ?:"},{"line_number":144,"context_line":"       id: parentId"},{"line_number":145,"context_line":"       type: Foo"},{"line_number":146,"context_line":"       dependsOnDestruction:"},{"line_number":147,"context_line":"         - id: childId"},{"line_number":148,"context_line":"           handler: onChildDestruction"},{"line_number":149,"context_line":"    child:"}],"source_content_type":"text/x-rst","patch_set":10,"id":"bacf61ea_877a4e75","line":146,"updated":"2016-08-05 21:46:39.000000000","message":"we decided to make it \"ralations\", aren\u0027t we?","commit_id":"ace35aca50be489ba10122cbbc907c7c0ac7880a"},{"author":{"_account_id":8127,"name":"Alexander Tivelkov","email":"locher@yandex.ru","username":"ativelkov"},"change_message_id":"3ae1f807263e18f8b2b57a5f4bcc6f2c2aa1c2f9","unresolved":false,"context_lines":[{"line_number":143,"context_line":"    ?:"},{"line_number":144,"context_line":"       id: parentId"},{"line_number":145,"context_line":"       type: Foo"},{"line_number":146,"context_line":"       dependsOnDestruction:"},{"line_number":147,"context_line":"         - id: childId"},{"line_number":148,"context_line":"           handler: onChildDestruction"},{"line_number":149,"context_line":"    child:"}],"source_content_type":"text/x-rst","patch_set":10,"id":"9ad45d7e_3af6e0d7","line":146,"in_reply_to":"bacf61ea_877a4e75","updated":"2016-08-08 13:03:28.000000000","message":"Yeah, we did. It got lost during some of the rebases here. Will restore it now","commit_id":"ace35aca50be489ba10122cbbc907c7c0ac7880a"}]}
