I took a different approach in a similar situation, dictated in large
part by the dynamic nature of my use-case (some dashboard components -
graphs, tables, etc - needed to be swapped in/out or moved on a daily
basis).
My solution was to wrap the components in "widgets" (ie. nested
dashboards) and build a dashboard generator using node.js. This tool
lets me specify the layout and reporting components using a
combination of a report specification (json) and a report definition
(sql tables) - when run, it generates the relevant .cdfde, .wcdf and
.cda files, wires all the dashboard, widget and component parameters
together* and generates a zip file ready for upload to a bi-server.
One potential benefit to this direction is (imo) the ability to
specify different "filter packages" and that the various dashboards
can select which set of filters they want to use at build-time. (I
had this tied in to complement a set of daily PDI transforms)
Best,
- Scott
* Although I've received blessing to share the tool, it's really not
ready yet. The parameter "auto-wiring" is still a bit buggy, a number
of features aren't supported yet and the project needs to be cleaned
up before any release. If there's interest, I'll setup a public repo
when I get a chance (likely not for several weeks... when will PDI let
me pause time? ;-)
>
https://groups.google.com/d/msgid/pentaho-community/CAC_LyUjhAhisH%2Bxm1psSvshE8Rdt32NreKJv-ZXNoQLpEcWWqg%40mail.gmail.com.
--
Scott Elcomb @psema4
http://psema4.com/pubkey.txt
http://www.pirateparty.ca/