This is an informal document. As it is mostly assembled from various notes and drawings, organization may come slowly. Likewise, it has not been thoroughly reviewed. I expect there to be mistakes, I expect to be ridiculed for them, but without the ridicule I cannot know to fix them. This experiment will be an outrageous failure without any feedback. My goals, as I imagine them at the moment, are these:
Present the analysis and design of a profiler data management and processing system.
Force myself to follow and document a more structured design methodology, and include any of my own comments on the process.
Attempt to share, inform, and encourage collaboration from the wider ATD developer community for at least the design and potentially the implementation. The more input others (developers have in the design, the more useful (and reusable) the final system will be for them.
Test the utility of the Unified Modeling Language (UML) in exchanging analysis and design information among developers and potential users. Likewise, test whether we can succeed in sharing any information at all and in sharing a "common vocabulary" for design and development efforts.
Experiment with the effectiveness of "release early, release often" in the design phase. The sooner design ideas are released for comment and revision, the sooner major mistakes, divergences, or incompatibilities are averted. Likewise, just as for software, incremental changes are easier to make than major ones.
Try out the Web as a medium for publishing and distributing design documents such as this.
Collect all of my notes and thoughts so far into something resembling organization which others can easily peruse.
From the above goals it should be obvious that I am hoping for lots of feedback, contributions, and collaboration. If something looks wrong, if you have a better idea on anything, if you have comments on the design or analysis process described, if you think my formatting is ugly, if you think my writing is bad or my spelling hopeless, please let me know. In the spirit of community development, everything in this document is revisable and up for debate. Most importantly, if you think a particular part of the system looks interesting to you, and you'd like to be the person who designs and implements it, then by all means step in. Only then will we really have an honest-to-goodness, open-source-development experiment.
This document should be available as HTML and as PostScript from my home page, with much credit due to the SGMLtools team:
I will try to update the following version string with each change in this document or in any of its supporting documents. I'll also try to include a log of changes between versions.
$Id: design.sgml,v 1.4 1999/06/14 20:10:52 granger Exp $
$Log: design.sgml,v $ # Revision 1.4 1999/06/14 20:10:52 granger # add scms html report from the framemaker source; add suds analyses in scms # skew-t archive # # Revision 1.3 1999/06/14 19:22:32 granger # finish additions from Erik interview; insert use case diagrams inline # # Revision 1.2 1999/06/05 01:43:17 granger # start documenting Erik's interview # # Revision 1.1 1999/05/12 18:47:33 granger # initial check-in of profiler design document and notes files # |