Sprint 1 Goal

Focus on update functionality with basic usage of AJAX, learning about different types of updates and row identification.

Front Burner Stories

As an Updater, I want to be able to refresh the result set.

Issues

• Support for sql script resources? (If macro body doesn't contain SQL statement)  Parameter mapping?  Transaction isolation? Heading row?
• Should there be an indication of what rows changed for this story?
• Integration testing phase does not have hooks for putting configuration into the atlassian-test-harness, or if it does, google doesn't know it.

Notes

• responsible use of AJAX, limit number of commits.

Update table with primary keys (no joins, one or more rows)

• Acceptance criteria:
  • varchars, ints supported
  • "clean" input accepted
  • exceptions accepted if user input is bad

Update table without primary key

• Acceptance criteria:
  • warning provided if multiple rows will be updated
  • cancel function provided
  • varchars, ints supported
  • "clean" input accepted
  • exceptions accepted if user input is bad

Design just came as extension of previous update story. Only needed to send old values for WHERE clause

Back Burner Stories

Update view of table with primary key

• Acceptance criteria:
  • varchars, ints supported
  • "clean" input accepted
  • exceptions accepted if user input is bad
• Issues:
  • Can we find out if there are primary keys?
  • Can we use JDBC MetaData?
  • Oracle vs. MySQL JDBC driver

Notes

There are issues with updating views of multiple tables.
The approach we think will work is to query for the view definition in vendor specific manor, (MySQL has INFORMATION_SCHEMA table, but slightly different for oracle). Allows for a mapping aliases in the view results to the select statement they map to.

E.g.:
Select C, A FROM view1;
View has columns A, B, C, D, where A maps to t1.c1, B to t1.c2, C to t2.c2, D to t2.c1
Note that the order of columns, and number is not exact from the select and view, and the view's columns could be from other views, meaning the mapping has to be semi-recursive.

Overall: View columns in the select are mapped to the real view columns, then to the table columns they come from.

Update using author provided information (column names, sql, ...)

- Note: "user provided information" will be refined during sprint. Currently considering providing a list of column names.

Notes

UI refactoring:

• Sorting columns
• In cell click for edit

TODO

*make user supplied rows uneditable
*where in callback does the old data replacement go

Update with complex datatype support

Edit

Issues:

• Updating a view without a primary key 
• Define author provided information

Charts

	Name	Size	Creator (Last Modifier)	Creation Date	Last Mod Date	Comment
	scrum-charts.xls	13 kb	Christopher Daniels	Jan 18, 2009	Jan 18, 2009	Initial version of story point burn up and task hours burn down charts for our first scrum of the sprint.	Edit | Remove

scrum-charts.xls (application/octet-stream)