A method and system for in-place interaction with embedded objects

Technical Field

This invention relates generally to a computer method and system for interacting with linked and embedded objects and, more specifically, to a method and system for editing and otherwise interacting with a contained object within the context of its container application.

Background of the Invention

Current document processing computer systems allow a user to prepare compound documents. A compound document is a document that contains information in various formats. For example, a compound document may contain data in text format, chart format, numerical format, etc. Figure 1 is an example of a compound document. In this example, the compound document 101 is generated as a report for a certain manufacturing project. The compound document 101 contains scheduling data 102, which is presented in chart format; budgeting data 103, which is presented in spreadsheet format; and explanatory data 104, which is presented in text format. In typical prior systems, a user generates the scheduling data 102 using a project management computer program and the budgeting data 103 using a spreadsheet computer program. After this data has been generated, the user creates the compound document 101, enters the explanatory data 104, and incorporates the scheduling data 102 and budgeting data 103 using a word processing computer program.

Figure 2 shows how the scheduling data, budgeting data, and explanatory data can be incorporated into the compound document. The user generates scheduling data using the project management program 201 and then stores the data in the clipboard 203. The user generates budgeting data using the spreadsheet program 204 and then stores the data in the clipboard 203. The clipboard 203 is an area of storage (disk or memory) that is typically accessible by any program. The project management program 201 and the spreadsheet program 204 typically store the data into the clipboard in a presentation format. A presentation format is a format in which the data is easily displayed on an output device. For example, the presentation format may be a bitmap that can be displayed with a standard bitmap block transfer operation (BitBlt). The storing of data into a clipboard is referred to as "copying" to the clipboard.

After data has been copied to the clipboard 203, the user starts up the word processing program 206 to create the compound document 101. The user enters the explanatory data 104 and specifies the locations in the compound document 101 to which the scheduling data and budgeting data that are in the clipboard 203 are to be copied. The copying of data from a clipboard to a document is referred to as "pasting" from the clipboard. The word processing program 206 then copies the scheduling data 102 and the budgeting data 103 from the clipboard 203 into the compound document 101 at the specified locations. Data that is copied from the clipboard into a compound document is referred to as "embedded" data. The word processing program 206 treats the embedded data as simple bitmaps that it displays with a BitBlt operation when rendering the compound document 101 on an output device. In some prior systems, a clipboard may only be able to store data for one copy command at a time. In such a system, the scheduling data can be copied to the clipboard and then pasted into the compound document. Then, the budgeting data can be copied to the clipboard and then pasted into the compound document.

Since word processors typically process only text data, users of the word processing program can move or delete embedded data, but cannot modify embedded data, unless the data is in text format. Thus, if a user wants to modify, for example, the budgeting data 103 that is in the compound document 101, the user starts the spreadsheet program 204, loads in the budgeting data 103 from a file, makes the modifications, copies the modifications to the clipboard 203, starts the word processing program 206, loads in the compound document 101, and pastes the modified clipboard data into the compound document 101. The spreadsheet program "implements" the spreadsheet data, that is, the spreadsheet program can be used to manipulate data that is in spreadsheet format. The format that a program implements is referred to as native format.

Some prior systems store links to the data to be included in the compound document rather than actually embedding the data. When a word processing program pastes the data from a clipboard into a compound document, a link is stored in the compound document. The link points to the data (typically residing in a file) to be included. These prior systems typically provide links to data in a format that the word processing program recognizes or treats as a presentation format. For example, when the word processing program 206 is directed by the user to paste the scheduling data and budgeting data into the compound document by linking, rather than embedding, the names of files in which the scheduling data and budgeting data reside in presentation format are inserted into the document. Several compound documents can contain links to the same data to allow one copy of the data to be shared by several compound documents.

Such prior art is described in Microsoft Windows Version 3.1 User's Guide, Microsoft Corp., USA, pages 487 to 514 and is reflected in the preambles of Claims 1 and 15.

Summary of the Invention

It is an object of the present invention to provide a method and system for interacting with a contained object within a window environment of a container application of a container object.

It is another object of the present invention to provide a method and system for combining menus of the container application with menus of a server application of the contained object.

These and other objects, which will become apparent as the invention is more fully described below, are provided by a computer method and system as claimed for interacting with a containee object contained within a container object. In a preferred embodiment, the container object has a container application with a container window environment that has container resources for interacting with the container object. The containee object has a server application with a server window environment with server resources for interacting with the containee object. The method of the present invention displays the container window environment on a display device. A user then selects the containee object. In response to selecting the containee object, the method integrates a plurality of the server resources with the displayed container window environment. When a user then selects a server resource, the method invokes the server application to process the server resource selection. Conversely, when a user selects a container resource, the method invokes the container application to process the container resource selection.

Brief Description of the Drawings

Figure 1 is an example of a compound document.

Figure 2 is a diagram showing how the scheduling data, budgeting data, and explanatory data can be incorporated into the compound document.

Figure 3 is a diagram of the sample compound document shown in Figure 1 as it appears when edited within the word processing application before in-place interaction occurs.

Figure 4 is a diagram of the embedded spreadsheet object as it appears when activated in place within the compound document.

Figure 5 is a diagram which shows the relationship between an object handler and the container and server processes.

Figure 6 is a block diagram of a sample instance of a linked or embedded object.

Figure 7 is a block diagram showing a public view of an object.

Figure 8 is a sample user menu provided by a container application to display and select the actions available for an object.

Figure 9 is a diagram showing the composite menu bar resulting from the merger of the server application menus with the container application menus of the example shown in Figure 4.

Figure 10 is a diagram of the menu groups that compose a composite menu bar in a preferred embodiment of the present invention.

Figure 11 is a diagram showing the component windows of a typical Single Document Interface application.

Figure 12 is a diagram showing the component windows of an Multiple Document Interface application.

Figure 13 is a block diagram showing the typical window hierarchy of a container application when it is editing an embedded object in place.

Figure 14 is a flow diagram showing message processing in an event-driven windowing operating system environment.

Figure 14B is a block diagram showing the public interfaces required to support in-place interaction.

Figure 15 is a flow diagram of an implementation of the IOLEInPlaceFrame::SetMenu method.

Figure 16 is a flow diagram of an implementation of the IOLEInPlaceFrame::EnableModeless method.

Figure 17 is a flow diagram of an implementation of the IOLEInPlaceParent::OnInPlaceActivate method.

Figure 18 is a flow diagram of an implementation of the IOLEInPlaceParent:: OnUIActivate method.

Figure 19 is a flow diagram of an implementation of the IOLEInPlaceParent::OnUIDeactivate method.

Figure 20 is a flow diagram of an implementation of the IOLEInPlaceObject::InPlaceDeactivate method.

Figure 21 is a flow diagram of an implementation of the IOLEInPlaceObject:: InPlaceUIDeactivate method.

Figure 22 is a flow diagram of an implementation of the IOLEInPlaceObject::Activate method.

Figure 23 is a flow diagram of an implementation of the ActivateUI function.

Figure 24 is a flow diagram of an implementation of the CreateObjectToolbars function.

Figure 25 is a block diagram of the shared menu data structure corresponding to the example discussed in Figure 4.

Figure 26 is a flow diagram of an implementation of the ObjectSetMenu function.

Figure 27 is a flow diagram of an implementation of the function Process_Object_Activation.

Figure 28 is a flow diagram of an implementation of the object linking and embedding API function ObjectLoad.

Figure 29 is a flow diagram of an implementation of the IOLEObject::DoVerb method. This method is the primary method for interacting with a containee object.

Figure 30 is a flow diagram of an implementation of the function Process_Activation_Message called by the window procedure of an MDI document window to process activation and deactivation messages.

Figure 31 is a flow diagram of an implementation of the Process_Mouse_LButtonUp function.

Detailed Description of the Invention

Table of Contents

<ul id="ul0001" list-style="none" compact="compact"><li>1. Overview</li><li>2. In-Place Interaction Overview</li><li>3. Window Support for In-Place Interaction</li><li>4. In-Place Interaction API <ul id="ul0002" list-style="none" compact="compact"><li>4.1 IOLEWindow Interface <ul id="ul0003" list-style="none" compact="compact"><li>4.1.1 IOLEWindow::GetWindow</li></ul></li><li>4.2 IOLEInPlaceUIWindow Interface <ul id="ul0004" list-style="none" compact="compact"><li>4.2.1 IOLEInPlaceUIWindow::GetBorder</li><li>4.2.2 IOLEInPlaceUIWindow::QueryBorderSpace</li><li>4.2.3 IOLEInPlaceUIWindow::SetBorderSpace</li></ul></li><li>4.3 IOLEInPlaceFrame Interface <ul id="ul0005" list-style="none" compact="compact"><li>4.3.1 IOLEInPlaceFrame::SetMenu</li><li>4.3.2 IOLEInPlaceFrames::InsertMenus</li><li>4.3.3 IOLEInPlaceFrame::RemoveMenus</li><li>4.3.4 IOLEInPlaceFrame::SetStatusText</li><li>4.3.5 IOLEInPlaceFrame::EnableModeless</li><li>4.3.6 IOLEInPlaceFrame::TranslateAccelerator</li></ul></li><li>4.4 IOLEInPlaceParent Interface <ul id="ul0006" list-style="none" compact="compact"><li>4.4.1 IOLEInPlaceParent::CanInPlaceDeactivate</li><li>4.4.2 IOLEInPlaceParent::OnInPlaceActivate</li><li>4.4.3 IOLEInPlaceParent::OnUIActivate</li><li>4.4.4 IOLEInPlaceParent::OnUIDeactivate</li><li>4.4.5 IOLEInPlaceParent::OnDeactivate</li><li>4.4.6 IOLEInPlaceParent::ShadeBorder</li><li>4.4.7 IOLEInPlaceParent::GetWindowContext</li></ul></li><li>4.5 IOLEInPlaceObject Interface <ul id="ul0007" list-style="none" compact="compact"><li>4.5.1 IOLEInPlaceObject::InPlaceDeactivate</li><li>4.5.2 IOLEInPlaceObject::InPlaceUIDeactivate</li><li>4.5.3 IOLEInPlaceObject::TranslateAccelerator</li><li>4.5.4 IOLEInPlaceObject::Activate</li><li>4.5.5 IOLEInPlaceObject::ResizeBorder</li><li>4.5.6 IOLEInPlaceObject::EnableModeless</li><li>4.5.7 IOLEInPlaceObject::SetVisRect</li></ul></li><li>4.6 Other Server Application Functions <ul id="ul0008" list-style="none" compact="compact"><li>4.6.1 ActivateUI</li><li>4.6.2 CreateNewMenu</li><li>4.6.3 CreateObjectToolbars</li><li>4.6.4 RemoveMenus</li></ul></li><li>4.7 Object Linking and Embedding API Helper Functions <ul id="ul0009" list-style="none" compact="compact"><li>4.7.1 SetActiveObjectHwnd</li><li>4.7.2 GetActiveObjectHwnd</li><li>4.7.3 ObjectCreateSharedMenu</li><li>4.7.4 ObjectDestroySharedMenu</li><li>4.7.5 ObjectShade</li><li>4.7.6 ObjectSetMenu</li></ul></li></ul></li><li>5. Use of In-Place Interaction API <ul id="ul0010" list-style="none" compact="compact"><li>5.1 Procedure for Activation in Place <ul id="ul0011" list-style="none" compact="compact"><li>5.1.1 Activation In Place Within a Multiple Document Interface Application</li></ul></li><li>5.2 User Selection of Pulldown Menus Message Handling</li><li>5.3 In-Place Deactivation Procedure</li><li>5.4 Closing the Container Application</li><li>5.5 Interacting with Modeless Dialogs</li><li>5.6 Handling Accelerator Key Combinations</li></ul></li></ul>

1.

Overview

The present invention provides a generalized method, referred to as in-place interaction, for interacting with embedded or linked data in the context of a compound document. That is, the application to be used to interact with the embedded or linked data is made accessible to the user through the windowing context (menus and windows) of the application that implements the compound document. This accessibility is referred to as activation in place. In a preferred embodiment, when embedded or linked (contained) data is activated in place, the menus of the application that implements the contained data are merged with the menus of the application that implements the compound document to create a composite menu bar. The order of the menus in the composite menu bar is determined by a set of menu groups. Each application categorizes its menus into these menu groups and places its menus in the composite menu bar in the order of the menu groups. The composite menu bar is then installed as the menu bar of the application implementing the compound document, and a message handler is installed to filter messages sent to the windows of this application. When the user selects a menu item, the message handler determines whether the menu item belongs to a menu of the application implementing the contained data or the application implementing the compound document. The message handler then sends the input message corresponding to the selected menu item to the correct application.

The present invention defines a set of abstract classes (interfaces) and functions through which contained data is activated in place. (In the C++ programming language, an abstract class is a class with a definition of its data and methods, but with no implementation for those methods. It is the responsibility of the application implementing the class to provide the actual code for the methods available to manipulate the class instance data.) The application implementing the compound document is responsible for implementing some of these interfaces and the application implementing the contained data is responsible for implementing others.

In a preferred embodiment of the present invention, an application program that creates a compound document controls the manipulation of linked or embedded data generated by another application. In object-oriented parlance, this data is referred to as an object. (The reference Budd, T., "An Introduction to Object-Oriented Programming," Addison-Wesley Publishing Co., Inc., 1991, provides an introduction to object-oriented concepts and terminology.) An object that is either linked or embedded into a compound document is "contained" within the document. Also, a compound document is referred to as a "container" object and the objects contained within a compound document are referred to as "contained" or "containee" objects. Referring to Figures 1 and 2, the scheduling data 102 and budgeting data 103 are containee objects and the compound document 101 is a container object. The user can indicate to the word processor that the user wants to edit a containee object, such as the budgeting data 103. When the user indicates that the budgeting data 103 is to be edited, the word processing program determines which application should be used to edit the budgeting data (e.g., the spreadsheet program) and launches (starts up) that application. The user can then manipulate the budgeting data using the launched application, and changes are reflected in the compound document. The same procedure is used whether the budgeting data is stored as an embedded or linked object.

If the application used to edit the budgeting data supports in-place interaction, then, when it is launched by the word processing program, it is activated within the window environment of the word processing program. Figures 3 and 4 illustrate the process of activating the embedded budgeting data 103 in place.

Figure 3 is a diagram of the sample compound document shown in Figure 1 as it appears when edited within the word processing application before in-place interaction occurs. The main window of the container application 301 contains a title bar 302, a menu bar 303, and a client window 304. The client window 304 displays the manufacturing project report discussed in Figure 1. The compound document contains an embedded spreadsheet object (the budgeting data 305). When the user edits the native text data of the compound document, the menu bar 303 appears as shown: it includes all of the commands necessary to interact with the word processing application.

When the user decides to edit the budgeting data 305, the user selects the spreadsheet object 305 and requests the word processing application to edit the object (e.g., by double clicking using the mouse). The word processing application then launches the spreadsheet application requesting that it edit the spreadsheet object 305. The spreadsheet application negotiates with the word processing application to edit the spreadsheet object 305 using windows 301 and 304 and the menu bar 303 of the word processing application.

Figure 4 is a diagram of the embedded spreadsheet object as it appears when activated in place within the compound document. The spreadsheet object 405 is edited directly in the client window 404 of the word processing application. The title bar 402 is changed to reflect that the application implementing the compound document, in this case a word processing application, is editing a spreadsheet worksheet within the compound document "VAC1.DOC." Also, the menu bar 403 is changed to a new composite menu bar, which comprises menus from the word processing application and menus from the spreadsheet application. In addition, various aspects of the embedded spreadsheet object 405 are changed to reflect that it is being edited within its container compound document. A selection highlight 406 in the form of a hatched border pattern is placed around the object. Also, the standard tools of the spreadsheet application, in this case the row and column markers 407, are placed around the spreadsheet object. Also, the spreadsheet selection cursor 408 is placed around the currently selected cell. At this point, the user is ready to edit the spreadsheet object 405 using all of the spreadsheet application commands.

In a preferred embodiment, application programs ("applications") cooperate using object linking and embedding facilities to create and manipulate compound documents. An application that creates a compound document is referred to as a container (or client) application, and an application that creates and manipulates containee objects is referred to as a server application. An application can behave as both a container and server. That is, an application can contain objects and the objects that the application implements can be contained within other objects. Referring to Figure 2, the project management program 201 and the spreadsheet program 204 are server applications, and the word processing program 206 is a container application. A container application is responsible for selection of the various objects within the container object and for invoking the proper server applications to manipulate the containee objects. Server applications are responsible for manipulating the contents of the containee objects.

In a preferred embodiment, applications are provided with an implementation-independent Application Programming Interface (API) that provides object linking and embedding functionality. The API is a set of functions that are invoked by container and server applications. These functions manage, among other things, the setup and initialization necessary for container applications to send and receive messages and data to and from server applications. The API provides functions to invoke server applications to manipulate containee objects.

The invoking of a server application can be relatively slow when the server application executes as a separate process from the container application. In certain situations this slowness may be particularly undesirable. For example, if a user wants to print a compound document that includes many containee objects, it may take an unacceptably long time to invoke the server process for each containee object and request each server process to print the object. To ameliorate this unacceptable performance, a server application can provide code that can be dynamically linked during runtime into the process of the container application to provide certain functionality more efficiently. This code is called an "object handler." Object handlers provide functionality on behalf of the server application so that the object linking and embedding API can avoid starting up server processes and passing messages to the server process. In the above example, an object handler could provide a print function that the object linking and embedding API could invoke to print a containee object.

Figure 5 is a diagram which shows the relationship between an object handler and the container and server processes. The object handler 502 is linked into the container process address space 501 during runtime by the object linking and embedding API 503. Typically, the object linking and embedding API 503 invokes the object handler 502 directly, and the container application code need not be aware that a handler is providing the functionality, rather than a server process 507.

In addition to providing a set of functions, the object linking and embedding ("OLE") API defines "interfaces" through which container applications can communicate with their contained objects. An interface is a set of methods (in C++ parlance) which abide by certain input, output, and behavior rules. If a contained object supports a particular interface, the container application can invoke the methods of that interface to effect the defined behavior. In a preferred embodiment, the container application does not directly access the object data. Rather, it preferably accesses the object data using the supported interfaces. A container application is bound to a contained object through a pointer to an interface. The container application accesses the object data by invoking the methods of the interface. To access the object data, the methods may send messages to the server application requesting the specified access. In a preferred embodiment, messages are sent between container and server applications when the server application is implemented as a separate process using interprocess communications mechanisms provided by the underlying operating system.

Figure 6 is a block diagram of a sample instance of a linked or embedded object. In a preferred embodiment, the layout of the instance conforms to the model defined in U.S. Patent Application Serial No. 07/682,537, entitled "A Method for Implementing Virtual Functions and Virtual Bases in a Compiler for an Object Oriented Programming Language". The instance contains object data structure 601 and interface data structure 613 for each supported interface. The object data structure 601 contains pointers 602 to the interface data structures 613 and may contain private data of the instance. The private data of this sample instance includes a class identifier 603, handle 604 to the storage for the object, and data 605 for tracking the state of the object. The class identifier (CLASS_ID) is used to access the appropriate server application for the object. It is similar to a data structure "type" used in programming languages. The interfaces can determine the server application for the object by using the CLASS_ID as an index into a persistent global registry. The persistent global registry is discussed further below. As shown in Figure 6, each interface data structure 613 contains a private data structure 606 and a virtual function table 608. The private data structure 606 contains a pointer 607 to the virtual function table 608. The virtual function table 608 contains pointers 609, 611 to the code 610, 612 that implements the methods of the interface. <tables id="tabl0001" num="0001"><img file="EP0820008A2_D0001.tif" /></tables>

Table 1 represents the definition for the interface for the first entry pintf₁ in the object data structure 601. In Table 1, the word "interface" is defined to mean a C++ class. The definition shows three methods with their parameters. The "=0" at the end of each parameter list indicates that the method has no code implementation. In the C++ programming language, these functions are termed "pure virtual functions". A class with a pure virtual function is referred to as an abstract class.

Figure 7 is a block diagram showing a public view of an object. The public view of an object is the various interfaces that the object supports 702-706. Each interface provides methods through which container applications can access the object. Each object supports an IUnknown interface 702. Container applications use the IUnknown interface 702 to determine which other interfaces the object supports. The implementation of IUnknown interface 702 for a particular object knows what other interfaces the object supports and returns to the invoking application pointers to those interfaces. In a preferred embodiment, the method IUnknown::QueryInterface is used for this purpose. Interfaces 703 through 706 are examples of typical interfaces that can be supported by an object. These interfaces derive from the IUnknown interface. For example, the IDataObject interface 703 provides methods for storing data in and retrieving data from the object. The IOLEContainer interface 704 provides methods for listing the containee objects that are contained within the object. The IPersistStorage interface 705 provides methods for storing the object to and retrieving the object from persistent storage. The IOLEObject interface 706 provides methods through which a container application invokes the functionality of an object that corresponds to a user-selected action.

In addition to the API, the object linking and embedding facilities of the present invention provide information to container and server applications through a persistent global "registry." This registry is a database of information such as (1) for each type of object, the server application that implements the object type, (2) the actions that each server application provides to container applications, (3) where the executable files for each server application are located, and (4) whether each server application has an associated object handler.

2.

In-Place Interaction Overview

Once objects have been linked or embedded into a document, a user can select objects and request that certain actions be performed upon the selected objects. A user requests actions by first selecting the object and then selecting an action (e.g., a menu item) to be performed upon the object. The implementing server application is then invoked to perform the selected action. One skilled in the art will appreciate that there are many ways to display the choices of possible actions to a user and allow the user to select an action. In a preferred embodiment, the container application determines from the global registry what actions are supported by the server application implementing the selected object and then displays the actions in a menu.

Figure 8 is a sample user menu provided by a container application to display and select the actions available for an object. Menu item 803 is the entry for the object on the container application Edit menu 802. The entry varies based on the currently selected object. When no embedded or linked objects are selected, menu item 803 is not displayed. Submenu 804 displays the actions supported by an "Excel Worksheet Object." In this example, the supported actions are "Edit," "Open," and "Type." The first action (e.g., "Edit") on a submenu is the default action, which is performed when a user double-clicks with a mouse pointing device on the object, or enters functionally equivalent keys.

Once a user has selected a desired action (from the menu or by double-clicking on the object), the container application can then invoke the server application passing it what action to perform on behalf of the container application. The container application does this by obtaining the IOLEObject interface for the object and then invoking the object's DoVerb method passing it the selected action. (The DoVerb method performs the object-specific actions on the object.) The server application in turn determines whether the object can be activated in place within the context of the container application. If so, the server application and container application merge their menus into a composite menu bar, negotiate the placement of server application tool bars, palettes, formula bars, etc., and set up merged message handling. At this point, the server application is ready to receive user input.

Continuing the example of Figure 4, the figure shows the user editing the spreadsheet object (the budgeting data 405) in place within the window environment of a word processing application. Figure 9 is a diagram showing the composite menu bar resulting from the merger of the server application menus with the container application menus of the example shown in Figure 4. The composite menu bar 901 comprises menus 902, 905 from the word processing application and menus 903, 904, 906 from the spreadsheet application. When the user selects a particular menu item from one of these menus, the container application through the merged message handler determines whether to dispatch the message to the word processing application or to the spreadsheet application.

In a preferred embodiment of the present invention, a composite menu bar is created based upon a set of predetermined conventions. Each application menu to be included in the composite menu bar is assigned to a menu group. The menus are then inserted into the composite menu bar according to the assigned menu group.

Figure 10 is a diagram of the menu groups that compose a composite menu bar in a preferred embodiment of the present invention. The composite menu bar 1003 comprises menu groups 1001 from the container application and menu groups 1002 from the server application. The container application menu groups 1001 include the File group, the Container group, and the Window group. The server application menu groups 1002 include the Edit group, the Object group, and the Help group. In a preferred embodiment, the container and server application menus are interleaved in the final composite menu bar, according to the Microsoft application user interface style guidelines, which is specified in "The Windows Interface: An Application Design Guide," Microsoft Corp., 1992, which is herein incorporated by reference. Specifically, in the composite menu bar 1003, the groups are arranged left to right in the following order: File, Edit, Container, Object, Window, and Help.

3.

Window Support for In-Place Interaction

In a preferred embodiment, the in-place interaction API is implemented using the capabilities of the underlying window system. The present invention is described assuming the underlying window system is similar to the Microsoft Windows 3.1 operating system ("Windows"), although one skilled in the art will appreciate that the present invention can be implemented in a different underlying window system. The Microsoft Windows 3.1 operating system is described in "Programmer's Reference, Volume 2: Functions," Microsoft Corp., 1992; "Programmer's Reference, Volume 3: Messages, Structures, and Macros," Microsoft Corp., 1992; and "Guide to Programming," Microsoft Corp., 1992, which are herein incorporated by reference.

In window environments, applications support a single document interface or a multiple document interface. A single document interface ("SDI") application interacts with one document (file) at a time. For example, a word processing application that supports SDI would display the file currently being edited in its primary window. A multiple document interface ("MDI") application interacts with multiple documents (files) by devoting at least one window to each document. For example, a word processing application that supports MDI would display each file currently being edited in a separate document window. The user selects the document window of the file the user wishes to edit either by clicking on the title bar of the desired document window or by selecting the window title from a list on the Window menu of the application.

Figure 11 is a diagram showing the component windows of a typical Single Document Interface application. A typical SDI application provides a frame window 1101, and, depending upon the application, may additionally provide pane windows 1105 and 1106 and a parent window 1107 for an embedded object resides. In the case of an SDI application, the frame window 1101 is also the document window. Pane windows 1105, 1106 provide multiple views of a compound document. A parent window 1107 may be created by the container application to delineate the object when the object is first inserted into the compound document. In the example shown in Figure 11, the embedded object is a spreadsheet object, which is displayed within an object window 1108, which is contained within the parent window 1107 of the container application. The object window 1108 is owned by the server application. The frame window 1101 contains a title bar 1102, a menu bar 1103, and a tool bar 1104. Typically, tool bars and other application-specific tools are attached to either the frame window or a pane window of a container application. They may also appear as floating palettes, which are windows that are independent of the windows shown in Figure 11 and thus appear to "float" on top.

Figure 12 is a diagram showing the component windows of a typical Multiple Document Interface application. A typical MDI application allows a user to edit multiple compound documents from within the same container application. In the example shown in Figure 12, the user edits two separate compound documents in the two document windows 1205, 1206. Each document window can contain pane windows in a manner analogous to the SDI application. Document window 1205 contains two pane windows 1207, 1208. Also, the MDI application can provide a parent window 1209 for containing embedded objects in a manner analogous to the SDI application. Figure 12 shows an embedded spreadsheet object presented within an object window 1210. As in the case of an SDI application, the application-specific tools may appear anywhere.

The windows managed by either an SDI or MDI application are created and maintained in a hierarchical fashion. Figure 13 is a block diagram showing the typical window hierarchy of a container application when it is editing an embedded object in place. The window hierarchy comprises container application windows 1301 from the container application, and server application windows 1307 from the server application. The container application 1302 manages its frame window 1303, which contains a document window 1304, which may contain a pane window 1305, which may contain a parent window 1306. When an object is activated in place, the server application 1308 creates a root window 1309 for the embedded object and any child windows it requires. The object root window 1309 contains object child windows 1310, 1311, 1312.

Every application, when implemented as a separate process, contains an input queue for receiving events connected with the windows residing in the application's window hierarchy. The window hierarchy of Figure 13 is supported by two different applications. Thus, there are separate input queues associated with the windows belonging to the container application and the windows belonging to the server application. Input queue 1313 is associated with the container application windows 1301. Input queue 1314 is associated with the server application windows 1307. When a user clicks with the mouse, or inputs keystrokes to one of these windows, the underlying window system puts an appropriate message on either the container input queue 1313 or the server application queue 1314.

4.

In-Place Interaction API

The object linking and embedding API provides functions and defines interfaces through which the container and server applications communicate to support in-place interaction. The methods of these interfaces and the other API functions are invoked by application code in the usual course of processing user input. In an event-driven windowing system, an application invokes the appropriate method or function in response to receiving a message indicating that a user has selected a particular menu item or object.

Figure 14 is a flow diagram showing message processing in an event-driven windowing operating system environment. Each window has its own message handler, which is registered with the underlying window system when the window is created. When messages are received on an application input queue (for example, the input queue of the container application 1313), the application filters, translates, or dispatches the message to the window system. The window system dispatcher in turn sends the message to the message handling function (the "message handler") that was previously registered for the particular window indicated in the message. Upon receipt of the message, the message handler processes the message. The processing may include using the object linking and embedding API. Steps 1401 and 1402 compose a message pump. In step 1401, the application waits for a message on its input queue. In step 1402, the application filters or translates the message, if appropriate, or dispatches the message to the windowing system dispatch function. Steps 1403 and 1404 are the steps in the window system dispatch function that dispatch the message to the appropriate window message handler. In step 1403, the window system dispatcher locates the message handler for the window that is indicated in the message. In step 1404, the window system dispatcher sends the message to the located message handler (e.g., by invoking the message handler).

Steps 1405 through 1412 compose a typical message handler for a window. A message handler is referred to as a "window procedure." In a preferred embodiment, if an application does not provide a window procedure for a particular window, the underlying window system provides a default window handler called DefWindowProc. In steps 1405 through 1408, the application decodes the message to determine what type of event has occurred. Typically, for each type of event, the application invokes a different function, as shown in steps 1409 through 1412. These functions may in turn use the object linking and embedding API. For example, when a menu event is received, the application, in step 1411, invokes a function that processes menu events. Step 1411 invokes the Process_Object_Activation function (shown as step 1413), which activates a containee object in place. As will be discussed further below, the Process_Object_Activation function uses the object linking and embedding API to activate a containee object.

The in-place interaction API defines the following interfaces: IOLEWindow, IOLEInPlaceUIWindow, IOLEInPlaceFrame, IOLEInPlaceParent, and IOLEInPlaceObject. The IOLEWindow interface provides a method for retrieving the window handle associated with one of the other interfaces. The IOLEInPlaceUIWindow interface provides methods through which a server application negotiates with a container application for placement of window tools. The IOLEInPlaceFrame interface provides methods through which a server application communicates with the frame window of a container application. The IOLEInPlaceParent interface provides methods through which a server application communicates with the parent window of a container application. The IOLEInPlaceObject interface provides methods through which a container application activates and deactivates a server application. Figure 14B is a block diagram showing the public interfaces required to support in-place interaction. The container object 14B01 supports the IOLEWindow interface 14B02, the IOLEInPlaceParent interface 14B03, and the IOLEInPlaceFrame interface 14B04. The containee object 14B05 supports the IOLEInPlaceObject interface 14B06. Each of these interfaces is described below in detail.

4.1

IOLEWindow Interface

Table 2 lists the IOLEWindow interface. In object-oriented parlance, the IOLEWindow interface is the "base class" for the other in-place interaction interfaces. Thus, the other interfaces are derived from the IOLEWindow interface and inherit its public methods. In the IOLEWindow interface there is only one public method called GetWindow. <tables id="tabl0002" num="0002"><img file="EP0820008A2_D0002.tif" /></tables>

4.1.1 IOLEWindow::GetWindow

The GetWindow method retrieves the window handle (unique window identifier) corresponding to the IOLEInPlaceUIWindow, IOLEInPlaceFrame, IOLEInPlaceParent, or IOLEInPlaceObject interface from which it was invoked. The retrieved window handle is typically used when invoking underlying window system functions.

4.2

IOLEInPlaceUIWindow Interface

Table 3 lists the IOLEInPlaceUIWindow interface. The IOLEInPlaceUIWindow interface methods are invoked by a server application to negotiate tool placement within the document and pane windows of a container application. <tables id="tabl0003" num="0003"><img file="EP0820008A2_D0003.tif" /></tables>

4.2.1 IOLEInPlaceUIWindow::GetBorder

The GetBorder method retrieves the location where the server application is allowed to place its tools (which are implemented as child windows). This method returns a rectangle located inside the frame of either a document or pane window, depending upon whether the document or pane interface was invoked. Once the server application has retrieved this rectangle, it can determine the width of space it needs to place any tools and can then request this space using the QueryBorderSpace method. If the rectangle returned by the container application is rejected, the server application can choose not to continue with activation in place or can choose to not activate its tools.

4.2.2 IOLEInPlaceUIWindow::QueryBorderSpace

The QueryBorderSpace method retrieves the designated amount of space in a pane or document window where the server application can place its tools. The method takes one parameter, a rectangle of border space, within the rectangle retrieved from a previous call to GetBorder, that the server application needs for its tool placement. The method returns an indication as to whether the document or pane window is able to accommodate the request. If the request cannot be accommodated, the server application can invoke the method again with a different rectangle, can choose not to continue activation in place, or can choose to not activate its tools.

4.2.3 IOLEInPlaceUIWindow::SetBorderSpace

The SetBorderSpace method informs its associated container application that it is actually allocating the designated space in the pane or document window to place the server application's tools. This method is called after the space has been successfully requested from the pane or document window in a previous call to QueryBorderSpace. The server application is responsible for allocating the space it needs. The method takes one parameter, the rectangle of space the server application is allocating to its tool child window. The designated rectangle may be smaller than that successfully previously requested. The term "designated" refers to a passed in parameter and "specified" refers to the interface, class, window, or object to which a particular method belongs. The method moves or sizes, as necessary, any of the specified pane or document window user interface resources.

4.3

IOLEInPlaceFrame Interface

Table 4 lists the IOLEInPlaceFrame interface. The IOLEInPlaceFrame interface provides methods invoked by a server application to communicate with the frame window of its container application. <tables id="tabl0004" num="0004"><img file="EP0820008A2_D0004.tif" /></tables>

4.3.1 IOLEInPlaceFrame::SetMenu

The SetMenu method installs and removes the designated composite menu bar as the menu bar of the container application and installs a message handler for the composite menu bar. Figure 15 is a flow diagram of an implementation of the IOLEInPlaceFrame::SetMenu method. This method uses different mechanisms to install the composite menu bar depending upon whether the container application is an MDI or SDI application. In step 1501, the method determines whether the designated composite menu bar is NULL and, if so, continues at step 1502, else continues at step 1503. In step 1502, the method invokes the helper function ObjectSetMenuDescriptor to remove the message handler for the composite menu bar, and returns. In step 1503, the method determines whether the container application is an SDI application, and if it is, continues at step 1504, else continues at step 1505. In step 1504, the method invokes the underlying window system function SetMenu to install the composite menu bar as the menu bar of the container application frame window, and then it continues at step 1507. In step 1505, the method sends a message to the frame window telling it to perform its MDI menu setup. In step 1506, the method invokes the underlying window system function DrawMenuBar to redraw the menu bar. In step 1507, the method invokes the helper function ObjectSetMenuDescriptor to install the message handler for the composite menu bar. In step 1508, the method performs any other processing that may be required at the time of changing its menu bar and then returns.

4.3.2 IOLEInPlaceFrame::InsertMenus

<tables id="tabl0005" num="0005"><img file="EP0820008A2_D0005.tif" /></tables>

The InsertMenus method inserts the menus of the container application into the composite menu bar being created by the server application. Code Table 1 shows pseudo-code for an implementation of the IOLEInPlaceFrame::InsertMenus method. The method takes two parameters: a composite menu bar and an array of menu counts. For each of the menu groups represented by the menus of the container application, there is a loop which inserts the menus for that group. In lines 1-10, if there are any File group menus, then the method inserts these menus in the composite menu bar, and increments the menu count array at the index corresponding to the File group. (For example, index=0 if the menu bar presented in the example of Figure 4 is used.) In lines 11-20, if there are any Container group menus, the method inserts these menus into the composite menu bar and increments the menu count array at the index corresponding to the Container group. Finally, in lines 21-30, if there are any Window group menus to be added, the method inserts these menus into the composite menu bar and then increments the menu count array at the index corresponding to the Window group. At the completion of this method, the value stored at each index in the menu count array indicates the number of menus that the container application inserted for that particular menu group. The method invokes standard functions from the underlying window system (CreateMenu and InsertMenu) to create the menus for the container application and to insert them in the composite menu bar.

4.3.3 IOLEInPlaceFrame::RemoveMenus

The RemoveMenus method allows the container application to remove its menus from the composite menu bar before the server application deallocates the composite menu bar. This method is invoked from the IOLEInPlaceObject::InPlaceDeactivate method. The RemoveMenus method takes one parameter: the handle of the composite menu bar where the container menus are stored. The composite menu bar is expected to be clear of all server menus before this method is invoked.

4.3.4 IOLEInPlaceFrame::SetStatusText

The SetStatusText method allows the server application to set the status window (if there is one) of the container application's frame window. Typically, the status window is located at the bottom of the frame window and contains status or hinting information corresponding to the current selection. The SetStatusText method is container application specific and will perform whatever operations the container application usually performs to set its status window. The method takes one parameter: the text string to insert in the status window.

4.3.5 IOLEInPlaceFrame::EnableModeless

The EnableModeless method enables or disables the currently displayed modeless dialog for the container application. A modeless dialog is an input window which is displayed until it is explicitly closed by the user. While this window is displayed, the user is able to interact with other windows. A modal dialog, on the other hand, is an input window which blocks out other window processing until the user enters acceptable input. This method is invoked by a server application when it wants to display a modal dialog, but and its associated container application is already displaying a modeless dialog.

Figure 16 is a flow diagram of an implementation of the IOLEInPlaceFrame::EnableModeless method. This method hides the modeless dialog of the container application, and when called again restores the modeless dialogs. If the designated flag fEnable is true, then the hidden dialogs are displayed, otherwise any currently displayed modeless dialogs are hidden (removed from display, but in-memory data structures not deallocated). In step 1601, the method determines whether FEnable is true, and, if it is, continues at step 1602, else continues at step 1603. In step 1602, the method invokes an underlying window system function ShowWindow to restore the windows associated with the previously saved modeless dialogs, and then returns. In step 1603, the method saves the window handle of the next currently displayed modeless dialog. In step 1604, the method invokes an underlying window system function ShowWindow to hide the window associated with the modeless dialog. In step 1605, the function checks to see if there are any more modeless dialogs displayed, and, if there are, the function loops back to step 1603, otherwise the function returns.

4.3.6 IOLEInPlaceFrame::TranslateAccelerator

The TranslateAccelerator method allows a container application to process accelerator key combinations when a server application receives a keystroke it does not recognize. Accelerator key combinations are keyboard shortcuts for menu commands and are discussed further below. The TranslateAccelerator method is invoked indirectly by the function ObjectTranslateAccelerator which is called in the server application message pump. The container application should perform its normal accelerator processing and then return an indication of whether or not the accelerator was processed. This value is then passed on to the server application by the function ObjectTranslateAccelerator. Note that because the message has been transferred from the server application to the container application, the underlying window system may not have retained any additional key state or message information associated with the designated message.

4.4

IOLEInPlaceParent Interface

Table 5 lists the IOLEInPlaceParent interface. The IOLEInPlaceParent interface provides the methods invoked by the server application to communicate with the parent window. This window is also referred to as the in-place "container site" for the object. <tables id="tabl0006" num="0006"><img file="EP0820008A2_D0006.tif" /></tables>

4.4.1 IOLEInPlaceParent::CanInPlaceActivate

The CanInPlaceActivate method is used by the server application to determine whether the container application supports in-place interaction. This method gives the container application a chance to accept or refuse the activation in place of a selected containee object. The method returns an indication of whether the container application is allowing in-place interaction.

4.4.2 IOLEInPlaceParent::OnInPlaceActivate

The OnInPlaceActivate method is invoked by a server application to give its container application a chance to perform any necessary operations before the server application creates the new composite menu bar (at activation time). Figure 17 is a flow diagram of an implementation of the IOLEInPlaceParent:: OnInPlaceActivate method. In this implementation, the only operation performed is setting a flag in step 1701 indicating that a containee object has been activated. This information is used later, whenever the specified object's parent container object is asked to activate or deactivate. This flag tells the parent container application whether it needs to activate or deactivate an object contained within it (a nested object), instead of activating or deactivating its own user interface.

4.4.3 IOLEInPlaceParent::OnUIActivate

The OnUIActivate method removes all of the container application menus and tools in preparation for activation of a containee object in place. Figure 18 is a flow diagram of an implementation of the IOLEInPlaceParent::OnUIActivate method. The steps performed by the method depend on whether the container object is itself an object that has been activated in place. In step 1801, the method determines whether the container object has been activated in place. If it has not, the method continues at step 1802, else it continues at step 1803. In step 1802, because the container object is a top level container object (not activated in place), the method uses its normal procedure to remove the container application menus and any extraneous tools, and then returns. In step 1803, because the container object is also a containee object, the method retrieves the object's own IOLEInPlaceObject interface to access the methods that treat the container object as a containee object. In step 1804, the method invokes the activate method of the container object to deactivate itself. In step 1805, the method hides all of the container object's document and pane window level tools. In step 1806, the method invokes the shade border method of the parent container object to remove the in-place interaction user feedback from around the container object and returns. The container object's object window is actually deactivated at a later time (e.g., when the containee object deactivates).

4.4.4 IOLEInPlaceParent::OnUIDeactivate

The OnUIDeactivate method is invoked by a server application at the end of deactivating its user interface resources to allow its parent container application to either activate its own user interface or invoke its parent container application to allow the parent container application to activate its user interface. Figure 19 is a flow diagram of an implementation of the IOLEInPlaceParent::OnUIDeactivate method. This method provides two different behaviors depending upon whether the container object is itself a containee object or is a top level container object. In the former case, if this container is to become the new object activated in place, then its own user interface is activated, otherwise the container requests its parent container application to activate its user interface. In the latter case, the container application restores its user interface using normal procedures. In step 1901, the method clears the flag indicating that the container application has activated a containee object. In step 1902, the method determines whether the specified container object is a containee object, and if it is not, continues at step 1903, else continues at step 1905. In step 1903, the method sets the container application menus and its title bar using normal procedures, and continues in step 1904 to set the input focus to the desired window, and returns. The input focus is set to a particular window when that window is to receive keyboard input. In step 1905, the method examines the flag ABOUT_TO_ACTIVATE to determine whether the container object is about to become the activated object, and if it is not, continues at step 1906, else continues at step 1907. (The ABOUT_TO_ACTIVATE flag is set when the container application is selected by the user, e.g., in the Process_Mouse_LButtonUp function discussed in detail below.) In step 1906, the method invokes the IOLEInPlaceParent::OnUIDeactivate method of the container object to activate the user interface of the container application of the parent container object, and returns. In step 1907, the method invokes function ActivateUI to activate the user interface of the container application, and returns.

4.4.5 IOLEInPlaceParent::OnDeactivate

The OnDeactivate method is invoked by the server application to give its associated container application a chance to free any resources or set flags associated with the activated containee object before the containee object is fully deactivated. The method is invoked from the IOLEInPlaceObject::InPlaceDeactivate method of the containee object.

4.4.6 IOLEInPlaceParent::ShadeBorder

The ShadeBorder method draws or removes a hatched pattern border from around the selected, or about to be deselected, containee object. The hatched pattern border is used to give the user feedback that the containee object has been activated in place. This method can invoke the helper object linking and embedding API function ObjectShade to create the proper shading pattern. The method takes two parameters: a rectangle surrounding the object where the border should be placed and a set of flags. The set of flags indicates whether the border should be on (SHADEBORDER_ON = 1) or off and whether the border should be drawn in the same color as the text contained in the title bar of the active window (SHADEBORDER_ACTIVE = 1) or the same color as disabled text. The active window is the window that has input focus.

4.4.7 IOLEInPlaceParent::GetWindowContext

The GetWindowContext method returns the set of container application interfaces associated with a particular containee object. Specifically, it returns the following parameters: <ul id="ul0012" list-style="none" compact="compact"><li>pFrame, which is a pointer to an IOLEInPlaceFrame interface;</li><li>pDoc, which is a pointer to an IOLEInPlaceUIWindow interface;</li><li>pPane, which is a pointer to an IOLEInPlaceUIWindow interface;</li><li>lprectChildPosn, which is a pointer to the location where the associated IOLEInPlaceParent instance will display the object window of the object within the parent window; and</li><li>hAccelTable, which is a handle to the container application's accelerator table (described below).</li></ul>

These values are used by the server application to negotiate and handle activation and deactivation. This method creates and associates instances of these interfaces with the relevant frame, document, pane, and parent windows of the container application.

4.5

IOLEInPlaceObject Interface

Table 6 lists the IOLEInPlaceObject interface. The IOLEInPlaceObject interface methods are invoked by a container application to activate and deactivate a contained object. Some of these methods access contained objects in a nested fashion, through the containment hierarchy. Others access only the current active object, which is the containee object displaying the editing menus. An alternative implementation would split this interface into two others: one to access only the active object and another to access a containee object through the containment hierarchy. <tables id="tabl0007" num="0007"><img file="EP0820008A2_D0007.tif" /></tables>

4.5.1 IOLEInPlaceObject::InPlaceDeactivate

The InPlaceDeactivate method is invoked by a container application to completely deactivate a containee object after an "undo" operation no longer needs to access the containee object and before the container application closes. This method performs the final deallocation of any resources associated with the activation of the containee object in place. Figure 20 is a flow diagram of an implementation of the IOLEInPlaceObject:: InPlaceDeactivate method. The method first determines whether it has activated a (nested) object contained within it, and if it has, it invokes the nested object's InPlaceDeactivate method. Otherwise, the object deactivates itself. In step 2001, the method determines whether the specified object is also a container object and has activated a nested containee object. If it has, the method continues at step 2002, else it continues at step 2004. In step 2002, the method retrieves the IOLEInPlaceObject interface of the activated containee object (which the server application of the specified object has previously stored), and in step 2003 invokes IOLEInPlaceObject::InPlaceDeactivate method of the retrieved interface, and then returns. In step 2004, the method checks to see whether the specified object's user interface is still active, and if it is, continues at step 2005, else continues at step 2006. In step 2005, the method invokes the specified object's InPlaceUIDeactivate method to deactivate its own user interface, and then continues at step 2006. In step 2006, the method invokes a server application function Remove_Menus to remove the server application menus from the composite menu bar. In step 2007, the method invokes the IOLEInPlaceFrame::RemoveMenus method of the specified object's container object to allow the parent container application to remove its menus from the composite menu bar. In step 2008, the method invokes the object linking and embedding API function ObjectDestroySharedMenu to deallocate the structure for the composite menu bar. The ObjectDestroySharedMenu function is window system specific and invokes whatever underlying window system functions are necessary to deallocate structures associated with a composite menu bar. In step 2009, the method invokes the underlying window system function DestroyMenu to deallocate the composite menu bar structure and to deallocate the window associated with the specified object. Finally, in step 2010, the method invokes the IOLEInPlaceParent:: OnDeactivate method of the specified object's container object, and returns.

4.5.2 IOLEInPlaceObject::InPlaceUIDeactivate

The InPlaceUIDeactivate method hides all of the user interface elements associated with the specified object that has been activated in place. This method is invoked either by the container application when it processes the user selection of a different object or area within the compound document, or from the specified object's InPlaceDeactivate function if the specified object's user interface has not yet been deactivated (See Figure 20). Figure 21 is a flow diagram of an implementation of the IOLEInPlaceObject:: InPlaceUIDeactivate method. This method first determines whether the specified object is a container object and has activated a nested object, and if so invokes the nested object's InPlaceUIDeactivate function. Otherwise, the method hides its own user interface and informs its container application that it has deactivated its user interface. In step 2101, the method determines whether the flag indicating that a nested containee object has been activated in place is true, and if so, continues at step 2102, else continues at step 2104. In step 2102, the method retrieves the IOLEInPlaceObject interface for the activated nested containee object, and in step 2103 invokes the nested containee object's InPlaceDeactivate method, and returns. In step 2104, the method clears the flag ABOUT_TO_ACTIVATE to indicate that the user has selected a different object. In step 2105, the method invokes the specified object's Activate method sending it a parameter of false to request the method to deactivate. This method removes all of the specified object's user interface elements that were associated with the parent container application frame window. In step 2106, the method invokes the object linking and embedding API function SetActiveObjectHwnd to remove the specified object's IOLEInPlaceObject interface from its association with the parent container application document window. This means that if the container application is an MDI application, and if the user later selects this document window, the specified object will no longer be reactivated in place. In step 2107, the method uses an underlying window system function to hide any user interface elements belonging to the server application that were associated with the parent container application's pane or document window. In step 2108, the method invokes the IOLEInPlaceParent:: ShadeBorder method of the specified object to remove the hatched border pattern feedback from around the deactivating object. In step 2109, the method invokes an underlying window system function to hide the window associated with the specified object. Finally, in step 2110, the method invokes the IOLEInPlaceParent:: OnUIDeactivate method to allow the container application to install its own user interface, and returns.

4.5.3 IOLEInPlaceObject::TranslateAccelerator

The TranslateAccelerator method allows a server application to process accelerator key combinations before the container application has a chance to process them. Accelerator key combinations are keyboard shortcuts for menu commands and are discussed further below. In a preferred embodiment of the present invention, the object activated in place, by convention, processes accelerator key combinations first. The Translate Accelerator method is invoked by the container application in its message pump (see Code Table 9). The only operation required to be performed by this method is to invoke the underlying window system function TranslateAccelerator with the specified server application accelerator table. Such invocation is not necessary if the containee object is implemented by a separate executable process, because the separate process receives these key combinations in its own message pump and the container application never receives them. In that case, the TranslateAccelerator method will do nothing.

4.5.4 IOLEInPlaceObject::Activate

The Activate method activates or deactivates the user interface elements installed in the frame window of the parent container application depending upon whether the designated flag fActive is true or false. If called when an MDI document window is activated or deactivated, this method installs or removes the composite menu bar associated with the object activated in place and puts a hatched border pattern around the specified object if it is being activated. If called when the top level frame window is activated or deactivated, this method places a hatched border pattern around the specified object if it is being activated, otherwise removes it. In this case there is no need to activate or deactivate other user interface elements. Figure 22 is a flow diagram of an implementation of the IOLEInPlaceObject::Activate method. In step 2201, the method determines whether it has been called as a result of activating or deactivating the top level frame window or an MID (child) document window. If called as a result of activating or deactivating an MDI document window, the method continues at step 2202, else continues at step 2210. In step 2202, the method determines whether the specified object is to be activated, and if, it is not, continues at step 2203, else it continues at step 2206. In step 2203, the method invokes the IOLEInPlaceFrame::SetMenu method of the parent container object to remove the composite menu bar associated with activation of the specified object in place. In step 2204, the method hides any user interface elements installed in the parent container application frame window. In step 2205, the method invokes the IOLEInPlaceParent::ShadeBorder method of the parent container object to remove the hatched border pattern from around the specified object, and returns. In step 2206, the method invokes the IOLEInPlaceFrame::SetMenu method of the parent container object to install the composite menu bar as the menu bar of the associated frame window. In step 2207, the method sets the title bar of the frame window of the container application to indicate that the container application has activated the specified object. In step 2208, the method invokes an underlying window system function to display any frame level user interface elements. In step 2209, the method invokes the IOLEInPlaceParent::ShadeBorder method of the parent container object to draw a hatched border pattern around the specified object, indicating that it has been activated in place, and returns. In step 2210, the method determines whether the specified object is to be activated, and if it is, continues at step 2209, else it continues at step 2211. in step 2211, the method removes the hatched border pattern from around the specified object, and returns.

4.5.5 IOLEInPlaceObject::ResizeBorder

The ResizeBorder method is called by the container application to request the server application to resize the user interface tools the server application has placed within the pane or document windows of its parent container application. In response to invocation of this method, the server application should begin another tool placement negotiation loop with the container application using the QueryBorderSpace and SetBorderSpace methods of the interface instance associated with the pane or document window.

4.5.6 IOLEInPlaceObject::EnableModeless

The EnableModeless method enables or disables the currently displayed modeless dialog for the server application. Typically, this method is implemented in a manner analogous to the IOLEInPlaceFrame::EnableModeless method, which is discussed above with reference to Figure 16.

4.5.7 IOLEInPlaceObject::SetVisRect

The SetVisRect method is called by the innermost level container object to communicate the amount of the object actually visible. The visible (clipping) rectangle of the object may have changed, for example, due to border negotiation, scrolling, or sizing. The designated rectangle is the clipping rectangle and it is the server application's responsibility to resize the containee object window to the correct (clipped) visible size.

4.6

Other Server Application Functions

In a preferred embodiment, a server application provides the following set of functions: ActivateUI, CreateNewMenu, CreateObjectToolbars, and RemoveMenus. These functions are shared by multiple server application interfaces including IOLEInPlaceObject and IOLEObject.

4.6.1 ActivateUI

<img file="EP0820008A2_D0008.tif" />

The ActivateUI function is a function implemented by a server application to control activation of the designated containee object's user interface resources. This high level function activates the frame, document, and pane level user interface elements, draws the hatched border pattern around the object, and displays the composite menu bar. Figure 23 is a flow diagram of an implementation of the ActivateUI function. The function takes two parameters: a pointer to a document interface and a pointer to a containee object. In step 2301, the function gets the window handles for the designated document window and the designated containee object. In step 2302, the function invokes the object linking and embedding API function SetActiveObjectHwnd to set the designated document window's currently active object to a pointer to the interface of the containee object. This enables a container application implemented as an MDI application to activate the proper containee object when one of its document windows is selected by the user. In step 2303, the function invokes IOLEInPlaceObject::Activate method to activate the designated object. In step 2304, the function invokes the underlying window system function ShowWindow to display any user interface elements associated with the container application pane or document windows. In step 2305, the function determines the dimensions of a border or rectangle to surround the designated containee object, and, in step 2306, the function invokes the IOLEInPlaceObject::ShadeBorder method of the designated containee object to draw a hatched border pattern around the designated containee object using this rectangle. In step 2307, the function sets the input focus to the object window of the designated containee object. Finally, in step 2308, the function invokes the underlying window system function DrawMenuBar to redisplay the composite menu bar, and returns.

4.6.2 CreateNewMenu

<img file="EP0820008A2_D0009.tif" /><tables id="tabl0008" num="0008"><img file="EP0820008A2_D0010.tif" /></tables>

The CreateNewMenu function is a function implemented by a server application to manage the creation of a composite menu bar. This function allocates the structures associated with the composite menu bar, requests the container application to insert its menus, and inserts the server application menus. Code Table 2 represents an implementation of the CreateNewMenu function. In line 1, the function invokes an underlying window system function to create the data structure for the composite menu bar. In line 2, the function invokes the IOLEInPlaceFrame::InsertMenus method of the container application frame window to insert the container application menus into the composite menu bar. In lines 3-5, the function tracks the number of menus the container application inserted for each menu group. In lines 6-17, assuming the server application has Edit group menus, the function creates each Edit group menu and inserts it into the correct spot in the composite menu bar, keeping track of how many menus it inserted. The correct spot is calculated in line 9 by determining how many menus have already been inserted to the left. For the Edit group, this will be the number of menus the container application inserted as part of the Container Group plus the number of Edit group menus already inserted, plus one for the current insertion. In lines 18-30, 31-41, the function performs analogous steps for any menus belonging to the Object group and Help group respectively. In line 42, the function invokes the object linking and embedding API function ObjectCreateSharedMenu to create the data structure associated with message handling for the composite menu bar and, in line 43, returns a handle to this structure.

4.6.3 CreateObjectToolbars

<img file="EP0820008A2_D0011.tif" />

The CreateObjectToolbars function is a function implemented by a server application to negotiate between the server and container applications for any space needed for server application tools. Figure 24 is a flow diagram of an implementation of the CreateObjectToolbars function. Steps 2401 through 2408 are repeated as many times as necessary to create tools for the server application. In step 2401, the function determines whether there are more tool bars to create, and if not, returns, else continues at step 2402. In step 2402, the function invokes the IOLEInPlaceUIWindow::GetBorder method of a frame, document, or pane window of the container application (depending upon where the server application desires to place the tools) to begin the negotiation. In step 2403, the function invokes the IOLEInPlaceUIWindow:: QueryBorderSpace method of the desired frame, document or pane window to request a specific width of border space inside the rectangle returned by the previous call to the GetBorder method. These methods were discussed in more detail above. In step 2404, if the specific width cannot be accommodated, then the function continues at 2405, else it returns. An implementation can choose to renegotiate for a different amount of space by calling QueryBorderSpace as many times as desired with different values. In step 2405, the function invokes the IOLEInPlaceUIWindow::SetBorderSpace method with the space previously negotiated in step 2403. In step 2406, the function creates a child window of the desired frame, document, or pane window of the parent container application. In step 2407, the function draws the tool in the child window it created earlier, and then returns to the beginning of the loop.

4.6.4 RemoveMenus

<img file="EP0820008A2_D0012.tif" /><tables id="tabl0009" num="0009"><img file="EP0820008A2_D0013.tif" /></tables>

The RemoveMenus function removes the menus of the server application from the composite menu bar upon containee object deactivation. The method is invoked from the containee object's InPlaceDeactivate method. Code Table 3 shows an implementation of the RemoveMenus function. The function uses the information stored in the shared menu descriptor (see Figure 25), which contains the number of menus within each menu group, to remove all of the server application menus by invoking the underlying window system function RemoveMenu. Lines 3-5 remove the menus belonging to the Edit group, lines 6-8 remove the menus belonging to the Object group, and lines 9-11 remove the menus belonging to the Help group.

4.7

Object Linking and Embedding API Helper Functions

In addition to interface definitions, the object linking and embedding API provides a set of helper functions to be used by container and server applications. These functions include the following: SetActiveObjectHwnd, GetActiveObjectHwnd, ObjectCreateSharedMenu, ObjectDestroySharedMenu, ObjectShade, and ObjectSetMenu.

4.7.1 SetActiveObjectHwnd

<img file="EP0820008A2_D0014.tif" />

The SetActiveObjectHwnd function sets the currently selected object in an MDI application. Each MDI (document) window in an MDI container application has associated with it an object interface corresponding to the object activated in place, which was displayed when that MDI window last had input focus. If no object had been activated in place from within the MDI window, then the associated object interface is NULL. This mechanism enables an MDI window to activate the proper containee object when the MDI window later receives input focus, for example, when the user clicks with the mouse on the title bar of the MDI window. (When the user clicks on the title bar of some other MDI window, any in-place interaction associated with the first window disappears from the display.) The SetActiveObjectHwnd function takes two parameters: the window handle of the MDI (document) window and the IOLEInPlaceObject interface of the object currently activated in place. One skilled in the art will realize that there are many ways to associate the window handle with an object interface. In one embodiment, the function stores the object interface as a property of the document window using underlying window system functions. Note that this implementation requires storage for each MDI window active in the system. An alternative approach is to add a method to the document, pane, and frame window interfaces to keep track of the currently selected object.

4.7.2 GetActiveObjectHwnd

<img file="EP0820008A2_D0015.tif" />

The GetActiveObjectHwnd function retrieves the containee object to be activated in place when the MDI window receives input focus. This function returns the object interface that was previously stored using the SetActiveObjectHwnd function.

4.7.3 ObjectCreateSharedMenu

<img file="EP0820008A2_D0016.tif" /><tables id="tabl0010" num="0010"><img file="EP0820008A2_D0017.tif" /></tables>

The ObjectCreateSharedMenu function creates the shared menu data structure associated with the composite menu bar of an object activated in place. The function is invoked from the CreateNewMenu function when a containee object is activated. Code Table 4 shows an implementation of the ObjectCreateSharedMenu function. The function takes two parameters: a handle to a composite menu bar and an array of menu counts which contains the number of menus in each menu group. The function returns a handle to the newly created shared menu data structure. In line 1, the function allocates the memory required for the shared menu data structure. In line 2, the function saves the handle to the composite menu bar in this data structure. In lines 3-15, the function sets up the shared menu descriptor according to the information stored in the menu count array. This descriptor is used by the window procedure, when it receives menu commands, to determine whether to forward the menu command to the container application or to the server application.

Although one skilled in the art will recognize that there are different ways to maintain this information, in one embodiment, the descriptor stores at each index the number of the last menu contained in the menu group associated with that index. (The menus are numbered from 1 on the left.) Also, an indication of whether the menu group belongs to the container or server application is stored at each index. One skilled in the art will also recognize that any menu grouping scheme could be supported by passing additional parameters to indicate which application should be notified for which menu group. Using this descriptor arrangement, the window procedure can determine within which index a particular menu item selection falls by counting the number of menus up to and including the menu item selection and then comparing the menu number with the descriptor values to find the correct index. Once the index has been determined, the window procedure can retrieve the indicator that specifies whether a container or server application function should be invoked. This procedure is discussed further below in reference to Code Table 5.

Figure 25 is a block diagram of the shared menu data structure corresponding to the example discussed in Figure 4. The shared menu data structure consists of a pointer to the composite menu bar 2502 and a descriptor 2505 which contains the menu count for each menu group. The composite menu bar 2503 comprises the menus from the container and server applications 2504. Each element in the descriptor 2505 has a count field 2506 and a function field 2507. The count field 2506 indicates the number, starting from the left, of the last menu within a menu group. For example, the second menu group is the Edit group and contains only one menu. This menu 2503 is the second menu from the left in the composite menu bar; therefore, the count field 2509 contains the number 2. As another example, the fourth menu group is the Object group. This group contains five menus from the server application 2510. Therefore, the count for this menu group 2511 contains the number 7, since the seventh menu is the Macro menu which is the last menu in the Object group.

4.7.4 ObjectDestroySharedMenu

<img file="EP0820008A2_D0018.tif" />

The ObjectDestroySharedMenu function destroys the shared menu data structure built in a previous call to ObjectCreateSharedMenu. This function is invoked from the IOLEInPlaceObject::InPlaceDeactivate method of the activated containee object after the container and server applications have removed their menus from the composite menu bar.

4.7.5 ObjectShade

<img file="EP0820008A2_D0019.tif" />

The ObjectShade function is provided by the object linking and embedding API to create the hatched border pattern placed around objects activated in place. The hwndParent parameter is the window handle associated with the IOLEInPlaceParent interface of the activated (or to be activated) object. The lprc parameter is the rectangle in the parent window coordinates where the pattern will be placed. The grfState flags are identical to those described in the IOLEInPlaceParent::ShadeBorder method and include SHADEBORDER_ON and SHADEBORDER_ACTIVE.

4.7.6 ObjectSetMenuDescriptor

<img file="EP0820008A2_D0020.tif" />

The ObjectSetMenuDescriptor function sets up or removes the message handler for the composite menu bar. The function is invoked by the IOLEInPlaceFrame::SetMenu method of the associated container object of the activating containee object. Figure 26 is a flow diagram of an implementation of the ObjectSetMenuDescriptor function. It takes three parameters: the window handle of the frame window associated with the container application, the handle to the shared menu data structure returned by the ObjectCreateSharedMenu function, and the window handle of the object to be currently activated in place. If the handle to the shared menu structure is null, then the function removes the message handler for the composite menu bar, otherwise it sets up the message handler. In step 2601, the function determines whether the handle to the designated shared menu data structure is null, and if it is continues at step 2602, else continues at step 2603. In step 2602, the function invokes the underlying window system function SetWindowLong to remove the special message handler that was previously associated with the container application. In step 2603, the function removes the properties previously set up for the composite menu and then returns. In step 2604, the function sets a property on the frame window to store the shared menu data structure to be later used by the special message handler. In step 2605, the function sets another property on the frame window corresponding to the window handle of the activating object. In step 2606, the function uses the underlying window system function SetWindowLong to install the special message handler as the new window procedure for the frame window of the parent container application. The old window procedure is saved for later use in the property Old_Filter. (See, e.g., Code Table 5, discussed in detail below.) The function then returns.

5.

Use of In-Place Interaction API

The object linking and embedding API functions supporting in-place interaction are invoked to: <ul id="ul0013" list-style="bullet" compact="compact"><li>activate an object in place within an SDI or MDI container application,</li><li>process messages when a user selects a menu item from the container application composite menu bar,</li><li>deactivate user interface resources of an object activated in place when the user either chooses to activate a different object or chooses to restore the top level container application to its normal processing,</li><li>deactivate the in-place interaction resources for a server application when the container application no longer needs them,</li><li>enable and disable modeless dialogs when a server application displays a dialog and the container application is currently displaying a modeless dialog (or vice versa), and</li><li>process accelerator key combinations to distribute them between container and server applications.</li></ul>

5.1

Procedure for Activation In Place

As described earlier, once objects have been linked or embedded into a document, a user can select objects and request that certain actions be performed upon the selected objects. Returning to the example of Figures 3 and 4, if the user wishes to activate the spreadsheet object 305 in place, the user can either click twice with the mouse input device on the object presentation format or use the container application menus to select an action on the object. Figure 8 demonstrates one way the user can use menus to activate the spreadsheet object 305 in place. When the user selects the menu item "Excel Worksheet Object" 803 from the container application (the word processing application) Edit menu 802, and then selects any action from the "Excel Worksheet Object" submenu 804, the word processing application invokes the spreadsheet application to activate the spreadsheet object in place.

The process of activating the spreadsheet object 305 is accomplished in several steps. First, the window procedure for the frame window of the word processing application is invoked by the underlying window system in response to user selection of a menu item on the object action submenu 804. (See, e.g., Figure 14.) Second, when a menu event is received, the window procedure invokes the function Process_Object_Activation. (See, e.g., steps 1407, 1411, and 1413.) Third, the function Process_Object_Activation loads the data for the spreadsheet object 305 using the object linking and embedding API function ObjectLoad. Finally, the function Process_Object_Activation program invokes the DoVerb method of the spreadsheet object 305 to request the spreadsheet application to perform the selected action.

Figure 27 is a flow diagram of an implementation of the function Process_Object_Activation. The function loads the selected object and invokes its DoVerb method to perform the selected action. In step 2701, the function invokes the object linking and embedding API function ObjectLoad, passing it a pointer to the storage for the object, and an indication that it wants the IOLEObject interface. The function ObjectLoad returns a pointer to the IOLEObject interface of the loaded object. In step 2702, the function invokes the SetClientSite method of the containee object is hand the containee object a pointer to its associated parent containee object interface (pclientsite). In step 2703, the function invokes the loaded object's IOLEObject::DoVerb method, passing it the selected action, an indication of whether the action was selected by a double click, and a previously created IOLEClientSite interface for the object. The function then returns.

Figure 28 is a flow diagram of an implementation of the object linking and embedding API function ObjectLoad. This function creates the in-memory instance of an object, readies a server application for future interaction, and returns a pointer to the designated interface. The function takes three parameters: a pointer to storage where the object data is to be loaded from, an indication of the interface the caller desires to have returned, and a return pointer to the in-memory instance of the object. In step 2801, the function retrieves the CLASS_ID from the designated storage. In step 2802, the function uses the retrieved CLASS_ID to locate the code for creating an in-memory instance of this type of object. In a preferred embodiment of the present invention, an IOLECreate interface is provided by every server application to create in-memory instances of objects it implements. In step 2803, the function invokes the IOLECreate::CreateInstance method to create the in-memory structure for the object and returns a pointer to an IPersistStorage interface through which the persistent storage for the object is accessed. In step 2804, the function invokes the IPersistStorage::Load method, which loads the object data from the designated storage. In step 2805, the function invokes the IPersistStorage::QueryInterface method to retrieve the designated interface, and returns the retrieved interface.

Figure 29 is a flow diagram of a typical implementation of the IOLEObject::DoVerb method. This method is the primary method for interacting with a containee object. The method negotiates with the container application to perform in-place interaction and causes the server application's user interface to be activated. The method takes four parameters: the user-selected action, a pointer to the message structure received by the container application window procedure when the user selected the action, a pointer to the IOLEClientSite interface of the object, and a set of flags controlling the execution of the verb, for example, whether the server application should take the input focus upon invocation. In step 2901, the method invokes the IOLEClientSite::QueryInterface method to get the IOLEInPlaceParent interface for the specified object. In step 2902, the method invokes the IOLEInPlaceParent::CanInPlaceActivate method to determine whether the container application supports in-place interaction. In step 2903, if the container application does not support in-place interaction, the method continues at step 2904, else the method continues at step 2906. In step 2904, the method creates and displays a server application frame window, because the container application could not perform in-place interaction. In step 2905, the method continues its normal processing of the designated action, and returns. In step 2906, the method invokes the IOLEInPlaceParent:: GetWindowContext method to obtain the interfaces associated with the container application. In step 2907, the method calculates the size of the object window it needs to create to support in-place interaction with the specified object. In step 2908, the method determines whether the area returned by the IOLEInPlaceParent::GetWindowContext method necessitates scaling or clipping and whether the specified object can support this. If it can support the required size, the method continues at step 2909, else the method abandons in-place interaction and continues at step 2904. In step 2909, the method retrieves the window handle for the window corresponding to the IOLEInPlaceParent interface. In step 2910, the method creates a new window as a child of the window corresponding to the IOLEInPlaceParent interface to be used as the object root window. (See, e.g., item 1309 in Figure 13.) In step 2911, the method determines whether the user interface resources for the specified object are still available, that is, have been allocated but not yet deallocated. If the resources are available, the method continues at step 2913, else the method continues at step 2912. In step 2912, the method invokes the IOLEInPlaceParent:: OnUIActivate method to enable the container application to remove its user interface resources in preparation for activation of the specified object in place. In step 2913, the method invokes the IOLEInPlaceParent::OnInPlaceActivate method to allow the container application to record that it has activated a nested object in place. In step 2914, the method invokes the function CreateNewMenu to create the new composite menu bar (see Code Table 2). In step 2915, the method determines whether the specified object requires the activation of any additional user interface tools, and if it does, continues at step 2916, else continues at step 2917. In step 2916, the method invokes the function CreateObjectToolbars to negotiate the location of and to place the additional user interface tools of the specified object (see Figure 24). In step 2917, the method invokes the function ActivateUI, which causes all of the user interface resources of the specified object to be displayed (see Figure 23), and returns.

5.1.1 Activation In Place Within a Multiple Document Interface Application

The previous section discussed the activation of a containee object assuming the object was activated from an SDI container application. If, on the other hand, the object is activated within an MDI container application, which application by definition can interact with multiple compound documents at the same time, then activation and deactivation occurs whenever the document (MDI) window containing the object is activated or deactivated. The window procedure for the document window receives an activation message from the underlying window system whenever the user selects the window (for example, by clicking in the title bar of the document window). The window procedure for the document window will receive a deactivation message when the user then selects a different window. In response to these messages, the window procedure for the document window will invoke a function (e.g., Process_Activation_Message) to perform the activation and deactivation of the document window and any activated object contained within it.

Figure 30 is a flow diagram of an implementation of the function Process_Activation_Message called by the window procedure of an MDI document window to process activation and deactivation messages. In one embodiment, the document window handle is passed in as a parameter to the function. The function determines whether the window contains an object previously activated in place when the window was last active. If so, the function activates or deactivates that object, else the function activates or deactivates the document window in its normal fashion. In step 3002, the function performs its normal window deactivation procedures, and returns. In step 3001, the function retrieves the IOLEInPlaceObject object interface for the previously active contained object, if there is one, by invoking the function GetActiveObjectHwnd. In step 3002, the function determines whether the object interface is null. If it is null, then there is no previously active contained object and the function continues at step 3003, else the function continues at step 3004. In step 3003, the function performs its normal window activation or deactivation procedures such as setting up the document window tools and menus, and returns. In step 3004, the function determines whether the designated flag FActive is true. If the flag FActive is true, then the previously active contained object is to be activated and the function continues at step 3005, else the object is to be deactivated and the function continues at step 3006. In step 3005, the function invokes the IOLEInPlaceObject::Activate method of the retrieved interface requesting the previously activated object to activate itself, and returns. In step 3006, the function invokes the IOLEInPlaceObject::Activate method of the retrieved interface requesting the previously activated object to deactivate itself, and then returns.

5.2

User Selection of Pulldown Menus Message Handling

Once the user has activated an object in place, the user interacts with the object within the container application by selecting actions through the menu bar of the container application (which is the composite menu bar). Because some of the menus belong to the server application and others of the menus belong to the container application, the window procedure for the container application frame window must decide whether to send the menu input event to a function within the container application or within the server application. For this purpose, a special message handler is installed by the object linking and embedding API function ObjectSetMenuDescriptor when it is invoked by the server application to install the newly created composite menu bar. This special message handler becomes the new window procedure for the container application frame window once it is installed. Thus, all messages received by the container application that correspond to its frame window are thereafter routed first to the special message handler. This special message handler then decides to which application to route the message event received. <tables id="tabl0011" num="0011"><img file="EP0820008A2_D0021.tif" /></tables><tables id="tabl0012" num="0012"><img file="EP0820008A2_D0022.tif" /></tables>

Code Table 5 shows an implementation of the special message handler provided by the object linking and embedding API. Lines 3-54 implement a case statement based upon the type of message received. Lines 28-49 provide initialization to enable the handler to route the message properly when it receives a menu command message from the underlying window system. Lines 4-27 and lines 50-53 provide the basic routing mechanisms when a menu command message is received. In line 29, the handle to the shared menu data structure is retrieved from the properties of the container application frame window. In lines 32-37, the handler walks the entries of the composite menu bar trying to match the menu received in its input message. In the process, it keeps track of the number of menus encountered (line 36). Once the loop is exited, and if the menu is found, the variable count represents the number of the selected menu starting from the left. This count is then used in the loop contained in lines 43-48 to determine to which descriptor element the menu belongs. Specifically, the value stored at each index of the descriptor is checked, and if the menu number is less than or equal to that value and is greater than the value stored in the descriptor at the index to the left, then the correct descriptor element has been located. Once the correct element is known, the handler retrieves the indicator corresponding to whether a container or server application function should be called. This indicator is then saved in the variable saveMenuRoutine in lines 45 and 47. For the handler to properly process menu mneumonics (including system menu key sequences), the handler sets the input focus to the frame window when responding to the menu command message.

A menu mneumonic is a key sequence such as "Alt, -, F, N" which provides a way to access menus using a keyboard instead of using a mouse. Typically, a menu item has one letter designated as its unique mneumonic, e.g., underlined on Windows 3.1. In order to access the menu item, its unique mneumonic is appended to the mneumonics of its containing menus. The entire key sequence is prefaced by the system key, which informs the system that the user wishes to type in a menu mneumonic. Menu mneumonics pose a special problem when in-place interaction is implemented with certain underlying window systems. Specifically, under unmodified conditions, when a containee object has been activated in place, all keyboard input with the exception of the System Key (the "ALT" key on Windows 3.1) and other navigation keys is sent to the window procedure for the object root window (see item 1309 in Figure 13), because the object root window is given input focus when the object is activated. However, the underlying window system sends the System Key to the window procedure for the container application frame window, because this key is handled specially. Thus, the container application will never receive the key input corresponding to the selected menu item in the system key sequence. For example, if the menu mneumonic key sequence is ALT, -, m (for "move"), the "ALT" key press would be sent to the container application and the "-" and "m" key presses would be sent to the server application. To solve this problem, the input focus is temporarily set to the container application frame window while the frame window procedure is processing menu events. An example of this solution is shown in lines 5, 39, and 40.

In lines 4-14, when the handler receives a menu command, the handler first restores the input focus so that normal processing may continue. Next, the handler checks the indicator that was saved in SaveMenuRoutine. If it indicates the menu belongs to the server application, then the handler invokes the window procedure of the object window, otherwise it invokes the original container application window procedure. Specifically, in line 7, the handler gets the window handle of the object, which was stored as a property of the frame window of the container application. In line 8, the handler asynchronously posts a message to the object window forwarding the original message and the original parameters. In line 11, if the indicator did not specify the server application, then the handler retrieves the original window procedure of the container application frame window and in line 12 invokes this procedure with the designated message and parameters. The original window procedure of the container application frame window was saved in the call to ObjectSetMenuDescriptor, which installed the new window procedure.

In lines 15-27, when the handler receives other menu related messages, it dispatches the message appropriately to either the container or object application in the same manner provided in lines 6-13. In lines 50-53, if any other kind of message is received, then the old window procedure is retrieved and the message and parameters are forwarded to it.

5.3

In-Place Deactivation Procedure

When the user selects an area outside the object activated in place, the object is deactivated. In general, this behavior occurs when the user clicks with a mouse button in another area in the window of the container application, or in a different document window in the case of an MDI container application. Therefore, generally speaking the deactivation methods of the in-place interaction API are invoked from a function called by the container application to process a mouse button event.

Figure 31 is a flow diagram of an implementation of the Process_Mouse_LButtonDown function. This function processes the input event signaled by the receipt of a left button down message. One skilled in the art will recognize that such a function could be invoked upon receipt of any kind of input event, and that other events can be used to select and deselect. In step 3101, the function sets a flag indicating that the container application is about to activate. This flag is used in the sequence of nested deactivation invocations to ensure the correct user interface resources are displayed in the case of nested activations. In step 3102, the function checks the flag ACTIVATED_SOMEONE to determine whether an object has been activated in place, and if it has, continues at step 3104, else continues at step 3103. In step 3103, the function performs its normal left button down event processing, and returns. In step 3104, the function retrieves the IOLEInPlaceObject interface for the currently activated object. In step 3105, the function invokes that object's IOLEInPlaceObject:: InPlaceUIDeactivate method. The function then continues at step 3103.

As shown in Figure 21, the currently activated object's IOLEInPlaceObject::InPlaceUIDeactivate method will in turn invoke its IOLEInPlaceParent::OnUIDeactivate method to allow the container application to install its user interface resources. This deactivation will nest upwards until either the top level container, or the container of the object about to be activated, is reached. (See Figure 19.) For example, referring to Figure 4, if the embedded chart object 409 shown within the embedded spreadsheet object 405 were the currently activated object, and the user selected the spreadsheet object 405 to activate it in place, then the chart's IOLEInPlaceObject:: InPlaceUIDeactivate method would be invoked, which would in turn invoke the spreadsheet's IOLEInPlaceObject:: OnUIDeactivate method. This latter method would then install the user interface for the spreadsheet object as shown in Figure 4. On the other hand, if the user had clicked somewhere else in the compound document, then the spreadsheet's IOLEInPlaceObject::OnUIDeactivate method would invoke the compound document's IOLEInPlaceObject:: OnUIDeactivate method. This latter method would install the word-processing user interface because this object is the top level container object.

A container application may display scroll bars for scrolling the displayed container object vertically or horizontally. The scroll bars are part of the container window. If the container object has an activated containee object, then when the user clicks on an area outside the containee object, the containee object is deactivated. In a preferred embodiment, the container application upon receiving a scroll bar message does not deactivate the containee object. Rather, the container application effects the scrolling and ensures that the input focus stays with the containee object.

5.4

Closing the Container Application

Some time after a container application is no longer able to perform an undo operation, which would reactivate a previously activated object, and before the container application is closed by the user, the container application permanently deallocates the user interface resources associated with a previously activated object. To deallocate these resources, the container application invokes the IOLEInPlaceObject::InPlaceDeactivate method associated with the previously activated object. This method in turn deallocates the shared menu data structure and the menus associated with the composite menu bar. (See Figure 20 and associated text.)

5.5

Interacting with Modeless Dialogs

When the user is interacting with an object activated in place whose server application has displayed a modeless dialog, then, if the user wants to select a menu item from a container application menu that presents its own modal dialog, then the container application temporarily hides the server application modeless dialog. Modeless dialog boxes are hidden because a user may be confused seeing two dialog boxes displayed simultaneously and not understand to which box the input is routed because the server and container applications are meant to appear as one application. Also, modal dialogs are not programmed to avoid conflicts with other dialogs because, in traditional applications, the underlying window system prohibits input outside the modal dialog within a single application. With the use of in-place interaction, such conflicts are not automatically avoided because two applications cooperate to appear as one. Therefore, the applicants should cooperate to avoid conflicts between modal and modeless dialogs. For example, suppose the user has selected the "Find..." menu item on the Edit menu of a spreadsheet application, which results in the display of a modeless dialog by the server application. Now, suppose the user wishes to print out a part of the compound document, so the user selects the "Print..." menu item on the File menu, which belongs to the word processing (container) application. The word processing application hides the "Find..." dialog because both dialogs are preferably not displayed at the same time. To do this, the word processing application invokes the IOLEInPlaceObject:: EnableModeless method of the spreadsheet application to request it to hide any modeless dialogs. Then, after the container application has finished processing the "Print..." dialog, it invokes the EnableModeless method to redisplay the modeless dialogs.

The similar situation can occur where the server application needs to hide a modeless dialog of the container application. In this case the IOLEInPlaceFrame::EnableModeless method is used.

5.6

Handling Accelerator Key Combinations

In a preferred embodiment of the present invention, the underlying window system supports a concept referred to as accelerator key combinations to enable the user to invoke menu commands through keyboard shortcuts. An accelerator key combination is a sequence of keys assigned by an application to be equivalent to invoking a particular menu command. For example, the key sequence consisting of pressing the "CNTRL" key followed by the pressing the "N" key might translate to the menu command "New" on the "File" menu. In a typical system, accelerator key combinations are assignable by the user and need to be unique within an application.

In general, accelerator key combinations ("accelerators") are processed in the message pump of an application (see step 1402 in Figure 14). A typical message pump invokes an underlying window system function passing it an accelerator translation table and lets the window system determine to which menu item command the accelerator corresponds. The window system then sends the resulting menu command to the appropriate window procedure.

A problem can be encountered in the object linking and embedding in-place interaction API with respect to accelerators. First, because a server application can be implemented as an object handler within the process space of its container application, the container application should ensure that the server application has a chance to translate its own accelerators. Preferably, the server application is given priority over the container application in processing ambiguous application accelerators when the server application has been activated in place. Also, in the case where a server application is implemented in its own process space, it should pass on any accelerators it does not recognize to the container application.

To solve this problem, the message pumps of the container application and the server application are modified to allow each other a chance to translate the application accelerators. Code Tables 6 and 7 show changes to the server application message pump applicable when the server application executes as a separate process. Code Table 8 shows changes to the container application message pump applicable when the server application executes within the same process as the container application (as an object handler). <tables id="tabl0013" num="0013"><img file="EP0820008A2_D0023.tif" /></tables>

Code Table 6 shows an implementation of the changes to the message pump of an object activated in place. These changes allow the server application to give the container application a chance to translate application accelerators before the server application (a separate process) finally disposes of an incoming message. In line 2, the server application attempts to translate an accelerator using its own translation table (stored in the variable hAccel). In line 3, if this translation was unsuccessful, either because there was no accelerator to translate or because the accelerator was not found in the server application translation table, then the server application invokes a special object linking and embedding API function ObjectTranslateAccelerator. The ObjectTranslateAccelerator function determines whether the accelerator is desired by the container application and, if so, sends a message through a remote procedure call to the container application to request it to translate the accelerator. The remote procedure call mechanism, due to its synchronous nature, ensures that the container application will process the message and return before the caller (the server process) receives any more input. In lines 4-5, if the container application did not translate the accelerator, then the server application handles the input message in its normal fashion (filtering and then dispatching it). <tables id="tabl0014" num="0014"><img file="EP0820008A2_D0024.tif" /></tables>

Code Table 7 shows an implementation of the object linking and embedding API function. ObjectTranslateAccelerator. This function allows the server application to give the container application a chance to process accelerators. In order to avoid the pitfalls inherent in synchronous message handling (such as indefinite waiting). ObjectTranslateAccelerator checks first to see if the container application is interested in the accelerator before attempting to invoke the container application's TranslateAccelerator method. The container application's accelerator table is a designated parameter passed in by the server application. It is accessible to the server application through a call to IOLEInPlaceParent:: GetWindowContext. If the container application's TranslateAccelerator method is invoked, this function returns the value returned by the container application to the server application, so that the server application can dispose of the message properly. <tables id="tabl0015" num="0015"><img file="EP0820008A2_D0025.tif" /></tables>

Code Table 8 represents a typical implementation of a message pump of a container application that supports in-place interaction. These changes allow the container application to give the server application (executing within the same process as the container application) a chance to translate application accelerators before the container application finally disposes of an incoming message. In line 2, the code retrieves the currently active object window handle associated with the document window of the container application. In lines 4-8, if there is an active object window handle, then the code invokes the IOLEInPlaceObject TranslateAccelerator method corresponding to the object window handle to enable the server application to translate the accelerator key combination. In lines 9-11, if the server application did not translate the accelerator, or if no object is active, then the container application attempts to translate any accelerators using its own translation table (stored in the variable hAccel). In lines 12-13, if there was no recognized accelerator to translate, then the container application handles the input message in its normal fashion (filtering and then dispatching it).