On 23.05.2018 we have had a workshop where we discussed the proposed features. Even though the general feedback was positive and none of the features was rejected, there's still a lot of work to do to take the effort to the production-grade level.
Below is the proposed mind-map with possible further steps that we could take (the source XMind file is also attached).
Rough effort estimation
Below can be found a table with description and coarse estimates of the possible further steps (time-wise* and priority-wise**).
* (Time-wise the tasks are estimated relatively to Aleksandr Pchelintcev, i.e. if the assignee of the issue wasn't fiddling with the UI thoroughly lately, some time has to be added to get in the context.)
** (Priority-wise the tasks go from the brightest ... ... to the coldest ) Develop an alternative to JCR for browsing and form viewing/editing. See the synergy with the REST client (or use some other approach). Goal - to have easily configurable REST browser and REST-based forms
MGNLUI-4439
-
Getting issue details...
STATUS
13 Validate foundation, APIs before next major. Ship with new major of REST Client any time. The field hierarchy is now kept in the form binders, so technically it is available while form is active and could be injectable in the validators and/or fields themselves. Need to see how to better expose such hierarchy and how to address at least some of the issues related to the topic. Timebox to 13 SP, possible followed by smaller efforts.
DEV-920
-
Getting issue details...
STATUS
Enables:
MGNLUI-2542
-
Getting issue details...
STATUS
Ship 1-2 alternative ways to layout forms (the API's are in place, even some implementation has started)
DEV-983
-
Getting issue details...
STATUS
DEV-974
-
Getting issue details...
STATUS
DEV-976
-
Getting issue details...
STATUS
DEV-978
-
Getting issue details...
STATUS
DEV-977
-
Getting issue details...
STATUS
There are probably bugs in the new UI implementation, maybe there will be potential blockers even. I can't estimate that, but I'd add at least 2.5-3 weeks for that on top of the total estimate.New features
Name Description: Estimate Resurface? Delivery Rest integration No Cross-field validation support 13 No Before 6.0 because new APIs are introduced. Forms with alternative layout 8 Maybe DEV-983 Before 6.0 Validation within composite fields Re-use 'validation bubble' effort from the past, try to provide universal validation/description UI regardless of the layout 1 week (though be trickier) [optional] Master-detail sub-app Test the flexibility of the new UI framework views and combine browser and detail in one sub-app (should be as easy as binding the pre-created form to the current selection in e.g. tree view) 3-5 days [optional] Filterable columns in Grids The PoC solution contains column renderer which allows to filter tree grid by path. Consider productising such feature and see how would it be possible to incorporate it into configuration and implementation of the content views. 1-2 weeks Essential features
Name Description Estimate Chooser dialog 1-2 weeks Multi-value field UI 1 week Migrate more fields 3 weeks (can be parallelised tho) Port more column renderers 1-2 weeks (can be parallelised) Harden JCR browsing implementation (remaining DS configuration, full-text search support etc) 1-2 weeks Multi-selection support in browser ValueContext needs generalisation (to conveniently cover the single vs multi selection cases) 3-5 days Harden DS observation mechanism ContentChangeEvent replacement needs to be hardened: should be memory-leak safe, should be clear, corner-cases should be considered. Observation should also work for the case of the detail sub-app 1-2 weeks Re-instate previews in actionbar Re-add the feature (should be simple with ValueContext). Question the abstraction of ImageProvider. 3-5 days Shortcuts in the grids and forms Currently not ported over from the old implementation. Need to see also how to do it better. 3-5 days Port form-dialog Thanks to the view-contexts and view improvements it is probably possible to make a form dialog implementation that shares most of the code with the detail sub-app, still requires quite some work. 1-2 weeks Implement Vaadin 8 - based inline row editing Grid/TreeGrid has in-built editor API, which can play together with parts of the from framework improvements that we introduced (binders/propertysets etc). Add support for inline editing in Grid/TreeGrid 1-2 weeks Migration strategy
Name Description Estimate App descriptor migration Think of a strategy (special sub-app that takes an old descriptor and translates to the new or some hybrid descriptor). Consider migrating whatever is easily migrated (fields, columns etc). 2-3 weeks Legacy action support It seems to be fairly sufficient to expose the selection context a set of JcrNodeAdapter to make some UI actions work. See how many we can cover by this and what is needed to cover more. 1-2 weeks Configuration
Name Description Estimate Content type powered config Requires content type solution to be available. Would be nice to shave some config parts off the detail sub-app and form dialogs. 1 week (once content types are in) Apply the type references where possible New UI improvements do not make the descriptors any leaner as is (in order to not make the configuration less powerful). With type references for whatever possible we could mitigate this fact. 2-2.5 weeks Actions
Name Description Estimate Re-start async action effort Use Java 8 instead of Quartz. Try to ship the asynchronous-ness as a trait/mix-in of an action, not as some base abstract class 2 weeks Misc
Name Description Estimate Unit testing The effort is almost not covered by unit tests (the state of the API and implementation was fluctuating too much so far) 3 weeks (+ maybe more since we'll need even more tests before final release) UI tests See if it would be possible to re-use the UI tests for our own apps when they are ported to the new framework 2 weeks UI push support finally incorporate Vaadin's push support 1 week Unknown unknowables