6. Notes for AV and Exercise Developers

6.1. Configuration: Code lines and Internationalization

Any JSAV-based exercise or AV (either standalone or an in-lined slideshow) can be associated with a configuration file. This is a .json file whose default name is the same as the name of the container for an inlined slideshow, or the same as the standalone AV or exercise. Configuration files support sections for defining all strings (used for internationalziation support), mapping logical names to code lines (to support alternate programming language examples in JSAV code objects), and setting defaults for configuration parameters.

6.1.1. File Format

The file format needs to be documented. In the meantime, a good example is AV/Sorting/insertionsortAV.json.

6.1.2. Use Case: Standalone AV or Proficiency Exercise

Standalone AVs and proficiency exercises are embedded into the HTML page using an iframe. The size of the iframe is taken from the size of a div element with classname container.

Given a standalone AV with HTML file foo.html that contains a div with classname avcontainer and JavaScript file foo.js, the configuration file would normally be named foo.json. After creating a configuration object, the string interpreter and code interpreters will typically be created as follows:

av = new JSAV($(".avcontainer"));
// Load the config object with interpreter and code created by odsaUtils.js
var config = ODSA.UTILS.loadConfig(),
    interpret = config.interpreter,       // get the interpreter
    code = config.code;                   // get the code object
var pseudo = av.code(code);

6.1.3. Use Case: Inline Slideshow

The RST file will reference a given slideshow with a specified div name as:

.. inlineav:: fooCON ss

The corresponding JavaScript file that implements the slideshow will create the JSAV, config, interpreter, and code objects with code like:

var av_name = "fooCON";
// Load the config object with interpreter and code created by odsaUtils.js
var config = ODSA.UTILS.loadConfig({"av_name": av_name}),
    interpret = config.interpreter,       // get the interpreter
    code = config.code;                   // get the code object
var av = new JSAV(av_name);
var pseudo = av.code(code);

6.1.4. Use Case: No pseudocode object

Creating the code object is not necessary if no pseudocode object is used in the visualization. A shortened version is therefore as follows:

var av_name = "fooCON";
var interpret = ODSA.UTILS.loadConfig({"av_name": av_name}).interpreter;
var av = new JSAV(av_name);

6.1.5. Use Case: Shared JSON file

Occasionally you might have multiple slideshows that share code or strings, such that it makes sense for them to share a configuration file. In this example, consider a JavaScript file fooCON.js that contains the code for one or more slideshows, with shared configuration file fooConfCON.json. The associated RST file would contain a line for each slideshow, such as:

.. inlineav:: fooS1CON ss

Each slideshow implemented in fooCON.js would contain code similar to:

var av_name = "fooS1CON";
// Load the config object with interpreter and code created by odsaUtils.js
var config = ODSA.UTILS.loadConfig(
               {"av_name": av_name, "json_path": "AV/Topic/fooConfCON.json"}),
   interpret = config.interpreter,       // get the interpreter
   code = config.code;                   // get the code object
var av = new JSAV(av_name);
var pseudo = av.code(code);

6.1.6. Using the configuration

After creating the interpreter and code objects, they can be used to replace text strings and line numbers as follows:

// If the .json file has definitions for av_c2 and av_c3 in various languages:
av.umsg(interpret("av_c2"));
av.label(interpret("av_c3", {top: 10, left: 10}).show();
// If the .json file has definitions for codelines with the tag ``loop``:
pseudo.setCurrentLine("loop");

6.2. URL Parameters

The client-side framework provides functionality to support easy processing of URL paramters in any stand-alone AV or exercise. See parseURLParams() in the Client-side Development section.

Some URL parameters are considered a standard part of the system and have built-in support. These include the parameters to control the natural language, the programming language (see Internationalization Support), and the exercise grading options. AV/exercise developers can also implement support for their own ad hoc URL parameters.

When an AV or exercise is embedded in an OpenDSA module via the avembed directive, the URL parameters are controlled by the configuration process. However, if a third party wishes to call the stand-alone AV or exercise directly (perhaps someone will want to embed calls within their own HTML pages), URL parameters are invoked as follows.

<URL>?JXOP-code=java

This one directs the AV to display Java code.

<URL>?JOP-lang=fi

This one directs the AV to use Finnish for its text.

Proficiency exercises typically support various grading modes. These are documented in the Configuration section. The typical options are as follows:

<URL>?JXOP-feedback=atend
<URL>?JXOP-feedback=continuous&JXOP-fixmode=undo
<URL>?JXOP-feedback=continuous&JXOP-fixmode=fix

6.3. Equations

Within jsav.umsg() text, all math should be done using LaTeX format enclosed within $...$ (for inline expressions) or $$ ... $$ (for display equations). MathJax will automatically recognize the dollar sign markup, and it will automatically do the conversion from LaTeX format to HTML. The only peculiarity that you should need to worry about is that backslashes must be escaped by using two backslashes. So a typical math markup within an AV or slideshow might look like:

jsav.umsg("This takes $\\Theta(n)$ time.");

6.4. CSS

Anything related to visual element style that is static should be defined in a CSS file. For example, if a JSAV array is placed at a specific location that never changes, then this location should be defined within a CSS file for your AV or slideshow.

While the client-side framework should automatically resize the AVs iFrame, developers should set the default height and width of the AV to accommodate the maximum size of the AV (such as an optional code block). If the automatic resizing should fail, the exercise should still be useable even if it doesn’t look as nice.

Some styling aspects are dynamic. For example, over the course of a visualization, nodes in a tree might need to change color to emphasize the action being visualized. Looking at the JSAV manual, you will notice that most visual elements can be styled with a .css() method on the element. But in nearly all cases, we wish to avoid using that method. We prefer to use the .addClass() and .removeClass() methods to control dynamic element styling whenever possible. These methods will dynamically assign or remove a CSS class to the element in the DOM. You can define any necessary new class in your AV’s CSS file. But before doing so, you should first check to see if a suitable class already exists in the OpenDSA style file at lib/odsaStyle.css. Given that we have developed a lot of visualizations already, the odds are pretty high that whatever visual styling you want to do is semantically equivalent to something that we already support. If so, you should be using the same style definition. For example, if you have a cell in an array or a node in a tree that your AV is currently acting on, then you probably want to indicate this by styling it using mynode.addClass("processing") for a tree node object named mynode, or using myarray.(index, "processing") for array position index in JSAV array myarray.

6.5. “Stand-alone” vs. “Inline” AVs and Exercises

Structurally, there are two ways that we include AVs and exerices into a module. First is the “stand-alone” artifact, which has its own HTML pages. In principle, this might be anything with its own URL, though in practice we usually only include our own materials. This is done using the avembed directive (see avembed). When converted to HTML, the mechanism used is a standard iframe tag to include the artifact.

“Inline” AVs are usually either a JSAV diagram or a JSAV slideshow (a diagram is just a “slideshow” with no slide controls at the top). These are included using the inlineav directive (see inlineav). The avID is the container name for the AV. Of course, the final HTML page has to get access to the relevent JavaScript and CSS files. This is done by putting at the bottom of the .rst file an odsascript directive giving the path and name of the Javascript file (see odsascript). If a CSS file is used, then you put near the top of the .rst file (right after the avmetadata block) an odsalink directive giving the path and name of the CSS file (see odsalink). Our naming convention is that all inlineavs use container names that end in CON, and that the .js and .css files use the container name. Further, our convention is that each individual slideshow or diagram be in its own JavaScript file (though this is convention is violated on occasion if there are a lot of very short slideshow files in a given page).

The odsascript and odsalink directives do nothing more than map down to <script></script> and <link></link> tags, respectively, in the final HTML pages. Their purpose is merely to keep module authors from needing to use raw HTML code in an RST file.

When you embed multiple slideshows on the page (with inlineav), they will naturally share the same namespace, both for code and for CSS.

For code, this is not generally an issue, because it is our standard procedure to wrap all of our code in an “anonymous function”, and then reference the key identifier (the container div) by name. This is why you will always see (in any of our code that has been cleaned to our internal spec, which should be everything except perhaps code in the Development directory) something like the following:

$(document).ready(function () {
  var av_name = "insertionsortS1CON";
  ...
  var av = new JSAV(av_name);
  ...
});

This does the following:

  • document.ready makes it wait until everything is loaded

  • It is all wrapped in a function, so that its namespace will not conflict with other slideshows. That way, for example, the global variables for one slideshow (like av in this example) are separate from the other slideshows. (This actually causes a problem if you want to include functions from other .js files. See Encapsulation.)

  • Use of the container name (such as in the JSAV call) is why THIS code gets executed on THIS container instead of the OTHER .js files that you loaded on the page.

Each inlineav might need to set some CSS styling with the same name as other slideshows will use. You handle this by “qualifying” the relevant variable to the name of the div that contains it. Look for example at AV/Binary/BSTCON.css to see examples. Notice lines that look like:

#avnameCON .jsav.jsavtreenode {
  ...
}

This will make your styling changes on the tree nodes only affect that particular slideshow.

6.6. Slideshows

The text in slideshows should be complete sentences. Which means that nearly always, there should be a period at the end of the sentence. The only exception would be when a series of slides is building up a sentence, such as if one slide said “First we do this…”, and then the following slide replaced it with “First we do this, then we do that.”

6.7. Programming Exercises

OpenDSA now uses the Code Workout system for implementing small coding exercises. See documentation from the Code Workout project for information about creating these exercises. The OpenDSA GitHub organization includes a private repository that holds the source for the Code Workout exercises along with the associated test cases. To see this you will need permission from the OpenDSA project management team. Contact us at opendsa@cs.vt.edu for further information.