To all:
While reading through Sten Vesterli's book "Oracle ADF Enterprise Application Development - Made Simple" I noted a brief discussion of project estimation and methodologies:
---
In order to produce the bottom-up estimate, you ask people capable of performing each task how much effort (in hours or days) it will take to produce the necessary output. Some projects prefer to let several people do independent estimates, while other project methodologies like Scrum prefer team estimates using collaborative techniques like "planning poker" (see
http://www.planningpoker.com)
---
There is also some good advice about breaking tasks into 'estimable' units. I'm curious, however, what experiences people have had in implementation of ADF application using Scrum methodologies and planning poker. I worked on a project a few months back where the project architecture was largely JSF (with a sprinkling of ADF Rich Faces), but we did the planning poker methods of project planning, focused on breaking tasks down into no larger than eight hours worth of effort. In practice, this meant that the effort was focused on breaking pages into UI components and managed/backing bean methods - so a single complex page view would be comprised of several smaller tasks corresponding to page elements. (i.e. "creating a table view plus the model for the table in the backing/managed bean", etc.)
I have not yet been part of an "ADF alone" project that utilized planning poker or the eight hour rule - most of them have been traditional waterfall types of projects. I'm curious if anyone has feedback or experience using Scrum or Agile methodologies with ADF - I'm thinking about a few questions:
1) In breaking tasks into 'smaller components', the natural breaking point for an ADF application seems to be to break sections of functionality into Bounded Task Flows. Have most projects gone about estimating efforts for a "Bounded Task Flow" (which in practice could be rather large), or if you're using scrum and small hour rules, is there an estimation method that works better than the BTF level?
2) In the business component layer, estimating views and entities based on "known data models" seems fairly straightforward - but in the case of agile development there is often the desire to avoid "Big Design Up Front", which in practical terms means that data models aren't always 'settled'. Does anyone have good advice on how to go about updating business components for a changing database - or a method that may work better than others. (There's no "rake migrate" option in ADF…)
I'm curious to know what experiences that people have had with Scrum/Agile and ADF.
Thanks!
- Chad