On Fri, Dec 22, 2017 at 8:19 AM, Vlad Rozov <vro...@apache.org> wrote:
> I don't see more complexity in implementing a plugin compared to > implementing an application. Additionally, for the use case you mention, > plugin is a better option, as likely the behavior applies not to a single > application, but all applications in that environment. > It would be developing a plugin in addition to an application as opposed to doing something directly in the application. Also what they may want to do may not be general enough or have enough reuse to justify developing a plugin. Our users typically build applications and not plugins so I would say most who have this need would not build a plugin but would do this directly in the application. > Websocket gateway address is a configuration parameter. A DAG may change > depending on configuration parameters (presence of a gateway, hadoop > version/vendor, security being enabled or disabled), but it should not > change depending whether DAG is populated for a launch or to get an info. > The configuration does not prompt the user to return a different DAG and even with plugins some kind of configuration hint is needed. There is no formal definition of what the method should and shouldn't do and attempt to define the method to only construct a DAG and not do anything else is not only retrospective but restrictive, for example should I not be able to connect to a syslog server and log something. What you are saying on the behavior w.r.t the population of the DAG with varying properties and the like is good practice. Like I mentioned earlier, users have always had full control of what they want to do in populateDAG method and what DAG they want to return and the platform does not particularly care what DAG is returned. It does not enforce nor rely that DAGs returned by multiple calls to populateDAG be the same DAG. > > Thank you, > > Vlad > > > On 12/21/17 10:05, Pramod Immaneni wrote: > >> Asking users to create plugins for something they want to do in their >> application logic is to do things in an indirect and cumbersome way with >> an >> added level of complexity. I don't think users will elect to do that. >> There >> is a reason populateDAG and the operators give users the flexibility they >> do today to have any custom logic they want. populateDAG isn't only for >> returning a constant DAG for an application, the configuration that is >> passed today to populateDAG, apart from hadoop environmental properties >> that could be considered constant also includes a variable component, >> which >> is the user customizable configuration properties. There are already >> examples of applications that have used these properties to do something >> different. Apart from the properties, some attributes are also injected >> into the DAG in a deliberate fashion by the platform to provide user with >> these so they can create the dag accordingly. One example is a websocket >> gateway address. If this is present applications create a websocket output >> operator else they end up create a console or some other output operator. >> >> On Thu, Dec 21, 2017 at 8:27 AM, Vlad Rozov <vro...@apache.org> wrote: >> >> "Sometimes" is not a use case. Config is not a context. >>> >>> Without concrete use cases the proposed change is not well justified. >>> populateDAG() is supposed to populate DAG, not to record anything in an >>> external system. It was a design goal for plugins. >>> >>> Thank you, >>> >>> Vlad >>> >>> >>> On 12/20/17 02:23, Priyanka Gugale wrote: >>> >>> +1 >>>> Sometimes this context is required. We shouldn't change any default >>>> behaviour other than making this config available. >>>> >>>> -Priyanka >>>> >>>> >>>> >>>> On Wed, Dec 20, 2017 at 5:32 AM, Pramod Immaneni < >>>> pra...@datatorrent.com> >>>> wrote: >>>> >>>> The external system recording was just an example, not a specific use >>>> >>>>> case. >>>>> The idea is to provide comprehensive information to populateDAG as to >>>>> the >>>>> context it is being called under. It is akin to the test mode or >>>>> simulate >>>>> flag that you see with various utilities. The platform cannot control >>>>> what >>>>> populateDAG does, even without this information, in multiple calls that >>>>> you >>>>> mention the application can return different DAGs by depending on >>>>> any external factor such as time of day or some external variable. This >>>>> is >>>>> to merely provide more context information in the config. It is upto >>>>> the >>>>> application to do what it wishes with it. >>>>> >>>>> On Tue, Dec 19, 2017 at 2:28 PM, Vlad Rozov <vro...@apache.org> wrote: >>>>> >>>>> -0.5: populateDAG() may be called by the platform as many times as it >>>>> >>>>>> needs (even in case it calls it only once now to launch an >>>>>> application). >>>>>> Passing different parameters to populateDAG() in simulate launch mode >>>>>> and >>>>>> actual launch may lead to different DAG being constructed for those >>>>>> two >>>>>> modes. Can't the use case you described be handled by a plugin? >>>>>> >>>>>> Thank you, >>>>>> >>>>>> Vlad >>>>>> >>>>>> >>>>>> On 12/19/17 10:06, Sanjay Pujare wrote: >>>>>> >>>>>> +1 although I prefer something that is more enforceable. So I like the >>>>>> >>>>>>> idea >>>>>>> of another method but that introduces incompatibility so may be in >>>>>>> 4.0? >>>>>>> >>>>>>> On Tue, Dec 19, 2017 at 9:40 AM, Munagala Ramanath < >>>>>>> amberar...@yahoo.com.invalid> wrote: >>>>>>> >>>>>>> +1 >>>>>>> >>>>>>> Ram >>>>>>>> On Tuesday, December 19, 2017, 8:33:21 AM PST, Pramod >>>>>>>> Immaneni < >>>>>>>> pra...@datatorrent.com> wrote: >>>>>>>> >>>>>>>> I have a mini proposal. The command get-app-package-info runs >>>>>>>> the >>>>>>>> populateDAG method of an application to construct the DAG but does >>>>>>>> not >>>>>>>> actually launch the DAG. An application developer does not know in >>>>>>>> >>>>>>>> which >>>>>>> >>>>>> context the populateDAG is being called. For example, if they are >>>>>> >>>>>>> recording >>>>>>>> application starts in an external system from populateDAG, they will >>>>>>>> >>>>>>>> have >>>>>>> >>>>>> false entries there. This can be solved in different ways such as >>>>>> >>>>>>> introducing another method in StreamingApplication or more parameters >>>>>>>> to populateDAG but a non disruptive option would be to add a >>>>>>>> property >>>>>>>> >>>>>>>> in >>>>>>> >>>>>> the configuration object that is passed to populateDAG to indicate if >>>>>> >>>>>>> it >>>>>>> >>>>>> is >>>>>> >>>>>>> simulate/test mode or real launch. An application developer can use >>>>>>>> >>>>>>>> this >>>>>>> >>>>>> property to take the appropriate actions. >>>>>> >>>>>>> Thanks >>>>>>>> >>>>>>>> >>>>>>>> >>>>>>>> >>>>>>>> >
