How to create a task pane in Office using pure C++/COM?

燕鹏李 301

I am very interested in the technology behind office add-ins. I have learned a little about its history, and know about several technologies that can be used to develop add-ins.

Office.js
VSTO
COM Shim
COM

I don't have good knowledge of Shim, maybe I am wrong in making it a separate category. After getting the book on VSTO(the book was published in 2007), I tried how to develop an office add-in using VSTO. I also do some extra development, such as embedding the Chromium kernel, so that I can load web pages using Chromium, instead of the IE kernel(it doesn't have good support on the new web standards).

What makes me sad is that, for using VSTO, I have to install the new two libraries, one is the .net framework, and the other is vstor(vsto runtime). On some machines, I cannot install these libraries successfully, so I prepare to develop office add-ins(in my case, the Word add-in) using C++/COM. (I guess Office was constructed using COM, we can use pure C++ to develop an add-in not depending on other libraries.)

But the really hard part is that COM technology is not popular now (sorry for what is popular) and the method of how to develop office add-ins using COM is even lost! I cannot search for any useful information on this topic on the Microsoft website.

Everything needs a start, my aim is to create a task pane(I need this to load web pages later) using C++/COM, but now, I don't how to achieve this, so I ask this question here, wishing to get some help.

RLWA32 40,021 Reputation points

2022-04-15T16:46:09.69+00:00

Put Task Panes aside for now. First you need to become familiar with the basic mechanics of Office Add-ins written in unmanaged C++ using COM.

There is an old Microsoft sample that you can retrieve from Outlook 2010: Building a C++ Add-In. This should give you a starting point. How familiar are you with COM? Have you ever used ATL? Have you ever used COM from an application to automate an Office application?
燕鹏李 301 Reputation points

2022-04-16T00:44:05.583+00:00
No, I never use COM directly to develop Office-related applications. Recently, in order to complete this goal, I searched some web pages, ready to download the following books:

Inside COM by Dale Rogerson

Essential COM by Don Box

Developer's Workshop to COM and ATL 3.0 by Andrew Troelsen

ATL Internals Working with ATL 8 by Christopher Tavares, Kirk Fertitta

Understanding COM+

Understanding DCOM

Sams Teach Yourself ATL Programming

Essential IDL - Interface Design for COM by Martin Gudgin

COM IDL and Interface Design by Dr. Al Major

Designing and using ActiveX Controls by Tom Armstrong

Inside OLE by Kraig Brockschmidt

Beginning ATL 3 COM Programming Richard Grimes

Professional ATL COM Programming by Richard Grimes

Windows via C/C++

Programming Windows
I don't know if these books are enough to help me develop Office COM Add-Ins. The experience I just deal directly with COM is using DirectX.
RLWA32 40,021 Reputation points

2022-04-16T06:57:33.477+00:00

That is quite a reading list. You don't need to read all of them to learn COM fundamentals. I have "Inside Ole 2nd Edition" by Kraig Brockschmidt and "Inside ATL" by George Shepherd and Brad King in my library. I may add "Essential COM" by Don Box.

Since your experience with COM seems to be limited I suggest you start slow and learn the fundamentals. The ATL framework does a wonderful job of simplifying COM development because it provides the scaffolding which can be used to speed implementation. It is a powerful tool. However, for someone just starting out the framework support obscures what is going on under the hood and behind the scenes.

As far as COM and Office applications go, you should understand the IDispatch interface, dual interfaces, threading models, in-process vs. out-of-process servers and how to use type libraries. Of course that's not an all-inclusive list. Browsing Office application type libraries using the Windows SDK utility oleview.exe can be very helpful when writing add-ins or code that automates an Office application. It is also essential to understand the object model of the Office applications with which you intend to interact. This understanding can be obtained by reading Microsoft's object model documentation written for VBA or .Net.

Before you plunge into add-ins I suggest you write a few simple COM Servers and related client applications to get the feel of how the technology works.
RLWA32 40,021 Reputation points

2022-04-16T06:59:37.31+00:00

And for general COM knowledge there is always Microsoft's own documentation.

https://learn.microsoft.com/en-us/windows/win32/com/com-fundamentals
燕鹏李 301 Reputation points

2022-04-16T10:03:21.593+00:00

Thanks for your advice. Regarding the development of Office add-ins, I have only dealt with Word Interop in the past. I will follow your advice, the development of the Office COM add-in is not a matter of a while, and it needs to add a lot of preliminary knowledge. In addition, I also want to systematically study the programming ideas covered by today's popular technologies.

I didn't know it was possible to use COM to develop Office add-ins. Later, I came across an Office add-in that was not developed with VSTO, but a .dll. I realized that there are other technologies that can develop Office add-ins. But there is very little information that can be found, just know that I can use COM. I know very little about COM, so I put the idea on hold. Later, I read the source code of a game engine, which also uses the idea of COM (Every module is implemented as a COM component). I think it is time to learn COM, so that by the way, I might also achieve the purpose of developing Office add-ins with COM.
RLWA32 40,021 Reputation points

2022-04-16T11:17:56.49+00:00
Office application add-ins created in unmanaged code with C++ and COM are in-process COM servers (DLLs). The interaction between Office applications and such add-ins follows a general pattern -

The Office application sees that an add-in should be used by checking registry data.

The Office application instantiates the add-in COM object and queries it for the _IDTExtensibility2 interface.

Upon receiving an interface pointer for _IDTExtensibility2 the Office application uses its methods to communicate with the add-in at various times (e.g., startup, shutdown, etc.) and passes the add-in the interface pointer for the Application Object for the relevant object model.

After receiving the Application Object the add-in can then use the Office application object model to achieve its objectives.

The add-in can also receive events from the Office application so that it can react accordingly.
燕鹏李 301 Reputation points

2022-04-16T13:24:46.027+00:00
Thank you for pointing me to how COM add-ins work. Although I have developed a commercial application before, I have not explored the working process of the Add-in you mentioned above :-). At that time, I only focused on which functions Office can call back, which interop interfaces provide which functions, and the overall add-in architecture was not further explored. Although I don't know more details, I can understand most of what you wrote here.

For now, what I'm interested in the Office COM add-in:

Remove the .NET dependencies;

Runs on almost all versions of Office (I don't know if Office 2003 works, but almost no one uses 2003, almost all users are using
Office 2016, 2019 and Microsoft 365);

It's easy to publish without going through the Office store.

But I also learned that because the COM add-in is unmanaged, once it crashes it will also cause Word to crash. I'm not worried about this, the program crashes because there are errors in it, and at this time, careful coding is required.
RLWA32 40,021 Reputation points

2022-04-16T13:40:53.36+00:00

It can be difficult to write a single add-in that is compatible with several different versions of Office. Microsoft makes changes to Office with each version that can affect add-ins. Features are added and removed. The object model changes also. And sometimes Microsoft makes major changes such as migrating from command bars to the ribbon for UI and moving to fast shutdown for Outlook. In my opinion having lots of version specific code in a single add-in to accommodate a variety of Office versions is prone to error.
燕鹏李 301 Reputation points

2022-04-16T14:09:44.72+00:00

This is a problem. I originally intended to be able to write only one program for Office 2007 and all Office after that, but now this may not be realistic. But at least it should be possible to write only one program for Office >= 2013. I hope so :( I don't involve outlook, only word. Even worst-case scenario doesn't matter much.
RLWA32 40,021 Reputation points

2022-04-16T14:55:38.743+00:00

I was just pointing out some general version related issues that could be encountered. Whether or not you would be affected depends on the specifics of your particular objectives, design and implementation.
Minxin Yu 9,866 Reputation points Microsoft Vendor

2022-04-19T07:53:30.147+00:00

As your issue is more related to Office developement, I would suggest you post new threads on following forums.
Word add-ins developer community
Excel add-ins developer community
PowerPoint add-ins developer community
Thanks for your understandings.
燕鹏李 301 Reputation points

2022-04-20T02:16:38.99+00:00

Thanks, I'll post the next question on the relevant forum.

1 answer

Bruce (SqlWork.com) 55,041 Reputation points

2022-04-16T19:09:45.903+00:00

com addins are only for legacy office applications. Also they are only supported for the desktop windows versions. Though I would not be surprised if even this support is dropped.

As the cloud became more important, a new plug-in architecture was developed for office. Now plug-ins are developed with html/css and JavaScript. The actual exchange server can serve as the installer (store) , and they work cross platform.

https://learn.microsoft.com/en-us/office/dev/add-ins/develop/develop-overview

The Office team supplies react components for all the office style controls, and matching css for office look. If you pick another framework besides react, they have a standalone sass/css package. The office team tends to use react for the official plug-ins so they are a good choice. You just need node installed to develop.

React fabric ui (sample code builds task pane)

https://learn.microsoft.com/en-us/office/dev/add-ins/design/using-office-ui-fabric-react

Fabric ui sass

https://learn.microsoft.com/en-us/office/dev/add-ins/design/fabric-core

Not sure why you would invest time learning an obsolete technology. The time it takes to learn to write a simple com component (office add-ins require discovery also be coded), you could probably code the complete plug-in modern style.
Please sign in to rate this answer.
燕鹏李 301 Reputation points

2022-04-17T02:57:46.99+00:00

Using COM to develop Office add-ins is indeed not the mainstream today, and there is a lack of materials, and the development is also difficult. For me, it's not a big problem that it only supports desktop systems, because in my case, even if I use office.js to develop add-ins, it is still only used on the desktop. In my place, most people don't have access to Office Online.

I want to choose COM for the following reasons.

Office.js only uses the system's built-in browser kernel, I can't switch it to Chromium (at least on Win7, Win10), using COM can load the dynamic link library of Chromium at startup, so that when rendering web pages, you can render with Chromium.

Many people are still using Office 2007, Office 2010, these versions cannot use add-ins developed by Office.js

For the above Office versions, if you use VSTO, additional dependencies (.net & vstor) will be generated, and some users fail to install these dependencies.

In my place, there is alternative software for Office, that is WPS, which has always been a competitor of Office. Many people use this software to replace Office(Office : WPS $\approx$ 6 : 4). This software is much more powerful than LibreOffice, and it can completely replace Office. This software also doesn't support office.js, but it can run add-ins developed with COM and VSTO.

Once you have an add-in that works on most office versions (in my case, Word), you can get the content of the Word document, which in turn can be combined with natural language processing, and communicate with remote services.

RLWA32 40,021 Reputation points

2022-04-19T11:02:25.54+00:00

Whether you use managed code or unmanaged C++/COM for your add-in you should know that your add-in will be responsible for creating and managing Task Panes in the Office application. For example, opening multiple existing workbooks in Excel or multiple existing documents in Word. Your add-in will also need to manage Task Panes for new items such as a new workbook or a new document. This is a non-trivial task.

燕鹏李 301 Reputation points

2022-04-20T02:06:09.71+00:00

Yes, what you just said is exactly a mistake I made in the past. I'm actually sharing the task pane between multiple documents, which is obviously a problem. When opening a new document, what you see to the right of the new document is the contents of the previous document's task pane. I've left that company and this bug has not been fixed yet :( When developing future add-ins, I'll be careful to manage task panes for multiple documents.
Sign in to comment