NOTE: Some of the sections of this template may not apply to your package, e.g. there may be no user-visible UI elements for a component of the OpenACS Core. Furthermore, it may be easier in some circumstances to join certain sections together, e.g. it may make sense to discuss the data model and transactions API together instead of putting them in separate sections. And on occasion, you may find it easier to structure the design discussion by the structure used in the requirements document. As this template is just a starting point, use your own judgment, consult with peers when possible, and adapt intelligently.
Also, bear in mind the audience for detailed design: fellow programmers who want to maintain/extend the software, AND parties interested in evaluating software quality.
When applicable, each of the following items should receive its own link:
OpenACS administrator directory
Subsite administrator directory
Tcl script directory (link to the API browser page for the package)
PL/SQL file (link to the API browser page for the package)
Transaction flow diagram
This section should provide an overview of the package and address at least the following issues:
What this package is intended to allow the user (or different classes of users) to accomplish.
Within reasonable bounds, what this package is not intended to allow users to accomplish.
The application domains where this package is most likely to be of use.
A high-level overview of how the package meets its requirements (which should have been documented elsewhere). This is to include relevant material from the "features" section of the cover sheet (the cover sheet is a wrapper doc with links to all other package docs).
Also worthy of treatment in this section:
When applicable, a careful demarcation between the functionality of this package and others which - at least superficially - appear to address the same requirements.
Note: it's entirely possible that a discussion of what a package is not intended to do differs from a discussion of future improvements for the package.
For a given set of requirements, typically many possible implementations and solutions exist. Although eventually only one solution is implemented, a discussion of the alternative solutions canvassed - noting why they were rejected - proves helpful to both current and future developers. All readers would be reminded as to why and how the particular solution developed over time, avoiding re-analysis of problems already solved.
Although currently only a few package documentation pages contain a discussion of competing software, (e.g. chat, portals), this section should be present whenever such competition exists.
If your package exhibits features missing from competing software, this fact should be underscored.
If your package lacks features which are present in competing software, the reasons for this should be discussed here; our sales team needs to be ready for inquiries regarding features our software lacks.
Note that such a discussion may differ from a discussion of a package's potential future improvements.
No single design solution can optimize every desirable software attribute. For example, an increase in the security of a system will likely entail a decrease in its ease-of-use, and an increase in the flexibility/generality of a system typically entails a decrease in the simplicity and efficiency of that system. Thus a developer must decide to put a higher value on some attributes over others: this section should include a discussion of the tradeoffs involved with the design chosen, and the reasons for your choices. Some areas of importance to keep in mind are:
Areas of interest to users:
Performance: availability and efficiency
Reliability and robustness
Areas of interest to developers:
Here's where you discuss the abstractions used by your package, such as the procedures encapsulating the legal transactions on the data model. Explain the organization of procedures and their particulars (detail above and beyond what is documented in the code), including:
Problem-domain components: key algorithms, e.g. a specialized statistics package would implement specific mathematical procedures.
User-interface components: e.g. HTML widgets that the package may need.
Data management components: procedures that provide a stable interface to database objects and legal transactions - the latter often correspond to tasks.
Remember that the correctness, completeness, and stability of the API and interface are what experienced members of our audience are looking for. This is a cultural shift for us at aD (as of mid-year 2000), in that we've previously always looked at the data models as key, and seldom spent much effort on the API (e.g. putting raw SQL in pages to handle transactions, instead of encapsulating them via procedures). Experience has taught us that we need to focus on the API for maintainability of our systems in the face of constant change.
The data model discussion should do more than merely display the SQL code, since this information is already be available via a link in the "essentials" section above. Instead, there should be a high-level discussion of how your data model meets your solution requirements: why the database entities were defined as they are, and what transactions you expect to occur. (There may be some overlap with the API section.) Here are some starting points:
The data model discussion should address the intended usage of each entity (table, trigger, view, procedure, etc.) when this information is not obvious from an inspection of the data model itself.
If a core service or other subsystem is being used (e.g., the new parties and groups, permissions, etc.) this should also be mentioned.
Any default permissions should be identified herein.
Discuss any data model extensions which tie into other packages.
Discuss modifications which the database may undergo from your package. Consider grouping legal transactions according to the invoking user class, i.e. transactions by an OpenACS-admin, by subsite-admin, by a user, by a developer, etc.
In this section, discuss user interface issues and pages to be built; you can organize by the expected classes of users. These may include:
OpenACS administrators (previously known as site-wide administrators)
You may want to include page mockups, site-maps, or other visual aids. Ideally this section is informed by some prototyping you've done, to establish the package's usability with the client and other interested parties.
Note: In order that developer documentation be uniform across different system documents, these users should herein be designated as "the developer," "the OpenACS-admin," "the sub-admin," and "the user," respectively.
Finally, note that as our templating system becomes more entrenched within the OpenACS, this section's details are likely to shift from UI specifics to template interface specifics.
Under OpenACS 5.7.0, parameters are set at two levels: at the global level by the OpenACS-admin, and at the subsite level by a sub-admin. In this section, list and discuss both levels of parameters.
If the system presently lacks useful/desirable features, note details here. You could also comment on non-functional improvements to the package, such as usability.
Note that a careful treatment of the earlier "competitive analysis" section can greatly facilitate the documenting of this section.
Although a system's data model file often contains this information, this isn't always the case. Furthermore, data model files often undergo substantial revision, making it difficult to track down the system creator. An additional complication: package documentation may be authored by people not directly involved in coding. Thus to avoid unnecessary confusion, include email links to the following roles as they may apply:
The revision history table below is for this template - modify it as needed for your actual design document.
|Document Revision #||Action Taken, Notes||When?||By Whom?|
|0.3||Edited further, incorporated feedback from Michael Yoon||9/05/2000||Kai Wu|
|0.1||Creation||8/21/2000||Josh Finkler, Audrey McLoghlin|