INVU has successfully integrated its document management product with over 15 other commercial pieces of software. Initial integration facilities concentrated on the ODMA software interface according to the standard developed by the American AIIMS group of document management system manufacturers. However, after extensive research by INVU it was found that the partner base felt the burden of programming for this interface, which has an 80-page specification, too onerous. This, combined with expensive integration costs from other software manufacturers, meant INVU looked for an easier solution.

The path chosen was to develop the ability to link with off-the-shelf commercial software products. In doing this, INVU's partners and customers would not need to access or change the underlying code but merely to train on the application in question. Hence CodeFree integration was built. Whilst giving general users of INVU the ability to use CodeFree and integrate with their particular solutions, it has also given INVU the ability to quickly and easily generate bespoke solutions where needed to satisfy customer demand.

The CodeFree Integration Solution
INVU's CodeFree Integration was released in December 2000. This satisfied two main requirements. It allowed a reseller or customer to develop their own graphical user interface, that could then, without any extra programming overhead, drive the INVU document management software to select previously-stored documents, optionally display them, and return the values of selected attributes. The second requirement was to perform the same tasks for any set of windows selected from those displayed by a commercial software product, so that these wndows would effectively provide a GUI onto the INVU system. Technically, both requirements are satisfied by the same technique of Intercepting windows messages generated when mouse-clicks are detected inside windows. The windows involved are identified by their title strings and by those of their ancestors, which unlike Windows handles, do not vary from run to run. Individual controls like buttons and edit boxes within the windows are identified by the same method, or by their coordinate positions. Once a window has been identified and its handle obtained, any variable text it contains can be obtained, and used to address the INVU database. Similarly, return values of pre-selected INVU attributes can be written to edit boxes in the client program.

The technical details of the last two paragraphs remain completely hidden from the customer. They are aware only of a Training Phase where the trainer must mouse-click on the buttons, scrolling lists and edit boxes in the client windows, that they want to use to control the INVU accesses, associating each with an INVU attribute that they select from a scrolling list.

The trainer can optionally choose to display the selected document and can choose attributes for writing back to selected edit boxes in the client. There are additional facilities to allow the values of selected attributes to be fixed at training time, which is very useful in restricting access to a particular area of the INVU database. Also, the values of selected attributes can be carried across from values set during an earlier INVU access. This allows a set of attribute values to be collected from across multiple client windows, and then used together to select an INVU document. The attribute values from earlier windows in the sequence will persist whilst an attribute value from the latest window is varied between accesses.

At run-time, the pre-trained INVU access is triggered by mouse clicking on an existing button, piece of static text or edit box, and the document display and / or the return value display occurs within a very short period of time. Where multiple documents match the input attribute values, these documents are listed for selection within a scrolling list, together with their INVU attribute values. Alternatively, the matching documents can be displayed in variable-sized sets, using multiple viewer windows, and providing a button to display the next
set. It is even possible to suppress document display and write back selected attribute values of selected documents at up to 10 documents per second. This would permit the client program to loop round reading these values and transferring them into its own database.