Why use an agent? If you decide to set up a library that exercises discrimination in choosing images to send, you'll want an agent. For example, if the library sends .PNG files to new browsers and .GIF files to older browsers, or smaller, lower-resolution images to customers from Web domains across the other side of the planet, then the built-in URL commands cannot keep up with your needs.

The demonstration Media Library contains a simple Web agent for returning an image. It doesn't make any of the sophisticated image choices described above, but it does perform some other simple improvements over merely sending back a file. It does this by dynamically building, and sending back to the browser, HTML to load the image, rather than the image itself. In so doing, it can supply the height and width of the image, plus any "Alt Text" defined in the Notes library.

Alt Text was originally used to display some text for non-graphical browsers or for users who switched off loading graphics to speed up browsing. Modern browsers will also show this text as a pop-up, or "tool tip", when the mouse is passed over the image. Supplying the width and height of an image in the <IMG> or <EMBED> tag is useful in that the browser can more accurately lay out the page, before downloading all the images, delivering a smoother browsing experience. Below is a sample Web agent output.

For Java applets, the sample Web agent can build an <APPLET> tag with all the parameters; it will also allow you to reference Notes file attachments in the applet parameters. Whenever the special term %CODEBASE% is found in an applet parameter, the %CODEBASE% term is changed when generating the tag into the required URL to access the file within Notes (e.g., %CODEBASE%ticker.wav). The sample below shows a Web agent output for Java applets.

Since the agent returns dynamic HTML code rather than an image, it must be called in a SRC parameter that expects text or HTML (for example FRAMESRC, IFRAME, ILAYER or LAYER). Alternatively, it can be loaded from non-Notes pages using SSI (Server Side Include), and another free program, GETHTTP.EXE can be downloaded from http://www.rhizomatics.demon.co.uk/software/gethttp.exe to assist that. GETHTTP.EXE will retrieve any Web address from the command line, and return the header, body or entire Web page to the screen or a file.

SSI example
Lots of people are put off SSI by the seeming complexity, but for simple uses it's very easy. You need to do no more than name the Web page with a .SHTML rather than .HTML and add in a line such as the example below that denotes where you wish the generated code to appear. The seam is invisible; to any user who views the HTML source, it will appear as if the dynamic HTML was part of the original SHTML file.