Wei Yongming: MiniGUI's Nirvana Rebirth Road

Background of the paper

MiniGUI is one of the most prestigious open source Linux graphical user interface support systems. This article is the article of Wei Yongming's latest original discussion on MiniGUI's Nirvana rebirth on September 1, 2017.

About the Author

Wei Yongming received a bachelor's and master's degree in engineering from Tsinghua University. The MiniGUI project hosted by the company is one of the most well-known open source software projects in the early days. It has been widely used in embedded products such as feature phones and digital photo frames. Wei Yongming founded Beijing Feiman Software Co., Ltd. in 2002 and successfully implemented a dual-license business model around open source software. In October 2015, he co-founded Beijing Yubotu Technology Co., Ltd. and served as CEO. Wei Yongming has done a lot of work for the popularization and application of Linux and open source software in China, translating or compiling a number of technical works ("Linux Practical Tutorial", "Linux Device Driver", etc.); in 2013, given his work in the field of open source software Outstanding contributions were selected by the "China Open Source Software Promotion Alliance" as "outstanding contributions to open source software."

I went to Shenzhen for a business trip last week. I had a few former colleagues from Feiman to eat and drink together. Gao currently is currently working for a CNC system developer in Shenzhen. Since 2000, Mr. Gao has been promoting the smart TV solution of Linux + MiniGUI in TCL and Skyworth, and even developed his own browser based on MiniGUI. However, as everyone knows, smart TV products have gained market recognition and great development and popularity in the past ten years (2010). Among them, Android has technically guaranteed the success of products such as smart TV. At the time, the total push was not recognized by the product department. In fact, it was not related to technology, but the market was still immature. Everyone talks about these past events, and they are all embarrassed.

As an open source software growing in domestic soil, MiniGUI released its first version in 1999, matured around 2003, and gained a large number of applications between 2000 and 2010 when the embedded product type flourished. It is used in industrial control systems, feature phones, digital photo frames and other types of products, including Huawei's digital photo frame, Lianxin's feature phone solution, and RDA's feature phone solution. I established Feynman Software in 2002, and tried to use the open source and commercial approach (dual license mode) to operate MiniGUI and its products, which has achieved a certain degree of success, and there are many lessons during the period.

Since the publication of the above article, eight years have passed, and AI has become the focus of competition in the next high-tech field. In the operating system and basic software, Android has occupied the market share of all smartphones except Apple, but it does not completely replace all Linux-based smart devices and other more embedded real-time systems. There are still many customers in the market. Using MiniGUI to develop various types of smart devices; on the other hand, MiniGUI's competitor Qt was easy to survive, but still survived. The word "embedded system" that was fashionable in the past has been replaced by "Internet of Things". Huawei's liteOS and other RTOS (real-time operating system) products such as RT-Thread and SylixOS developed by local developers have obtained government departments and even capital markets. The support has thus gained new development opportunities.

As the original creator of MiniGUI, I don't have to say anything about this software. However, in addition to me, there are more Feiman software colleagues (including the early users and developers of these MiniGUIs) who have devoted a lot of efforts to the development of MiniGUI.

Wan is currently working for a star enterprise in Shenzhen. He was the head of the Shenzhen R&D team. Hou was an architect and now he is a pioneer in Zhuhai. In 2007, the Feiman Shenzhen team under the leadership of Wanhehou began to take charge of the maintenance of MiniGUI. Around 2009, the team developed a complete set of animation support frameworks and interface development techniques for MiniGUI. Without the birth of Android, these technologies could have been applied to the upgrade of feature phones, providing excellent UI/UX solutions for feature phones. But the open source of Android is free, and the promotion of Google has made these technologies lose the opportunity of application and development.

Dong currently works for Xiaomi and is the main developer of miniStudio products. miniStudio is an Eclipse-based integrated development environment that provides a "what you see is what you get" GUI interface editing tool. In order to realize the programming convenience of the GUI application, Dong modified the programming framework of the MiniGUI control, developed the mGNCS (new control set) component, and introduced an advanced UI customization system and solution such as the appearance renderer. The leadership team developed the miniStudio product. . Today, after many years, when I found the miniStudio 1.0.8 for Windows version from the server, when I installed it on my computer and running it, I was still amazed at the visual programming capabilities provided by this software, even today, it is very valuable! However, Feiman Software did not spend any effort to promote and develop miniStudio, nor did it develop a business model based on miniStudio that can be a virtuous circle.

mDolphin is a WebKit-based browser product. Zhang and Lu are the two main developers. The development of this product has been continuously evolving for many years, and even successfully applied to the functional mobile phone solution of Lianxin. mDolphin is a typical large-scale application software on MiniGUI. In addition to its own application value, it also strongly proves the maturity of MiniGUI products. At present, Zhang works as a technical expert in Alibaba, and Lu is a pioneer in the Internet field.

The former colleagues who have worked on Feiman Software mentioned above have paid their own youthful years for the development of MiniGUI, and their technical ability has also been greatly improved and exercised. Regardless of the industry in which they later worked, or working or starting a business, the development work done by FMSoft around MiniGUI is still an important part of its resume. In addition to them, in the development process of more than ten years, Feiman Software has trained nearly 100 outstanding senior software engineers for the industry, which cannot be listed one by one. Some of them work for key technical positions in large Internet companies such as Tencent and Ali, and some start businesses in the mobile Internet and other fields. They are dedicated to the development and maintenance of MiniGUI, which will benefit them for a lifetime.

When I went to Shenzhen last week, when I was able to demonstrate and tell the animation framework and related technologies on MiniGUI to his current colleagues, I was deeply impressed that Gao, Wan, Hou, Dong and others were MiniGUI. The developers who have devoted their development to the MiniGUI are no worse than me.

When Gao had a meal with them, Mr. Gao mentioned that MiniGUI still has strong demand and can continue to develop. In fact, on this issue, almost every time I go to Shenzhen and Gao, Hou, they eat and drink, they will come up to argue. They think that MiniGUI still has strong market demand and should continue to develop. It is a pity to leave it. I think that each product has its own life cycle. MiniGUI has been used for more than ten years since 1999. It’s almost over.

Since 2010, I have led the team of the original Feynman software to develop the Tool App (Leading Desktop, Lock Screen, etc.) on Android, customized the smart TV solution, and now start the business in the field of intelligent hardware and mobile Internet. Probably from 2012 onwards, the development of MiniGUI is in a state of stagnation. Except for the latest version developed by Wan and Hou, and the open source release of mDolphin, there has been almost no major update. However, since 2015, the demand for MiniGUI has become more and more vigorous. Every year, customers will take the initiative to find Feiman software to obtain the commercial license of MiniGUI.

After many years of tossing, I went back to reflect on the path that MiniGUI and Feiman software have gone through. The reason why I think that the life cycle of MiniGUI has arrived is probably due to the failure of Feiman Software to achieve success. In other words, I confused technology and business. A business may not be able to operate in a single mode, but the technology will not be easily outdated.

Give a typical example of MySQL. MySQL's early operating company used a dual-license model and achieved a certain degree of commercial success. Later, it was acquired by Oracle. The dual-licensing model has basically completed its historical mission, but the application of MySQL is related to the Internet, mobile Internet, With the development of cloud services becoming more and more popular, even after many NoSQL databases appeared, MySQL still showed its vitality.

For example, specifications or infrastructure such as OpenGL have been developed for decades and have never seen their obsolescence. Apache, X Window, and other technologies or projects are almost always the evergreen of the open source world, not to mention the Linux kernel.

Obviously, as long as it is the basic software with market demand and user base, continuous development is its main theme; it is business that is outdated, not technology itself.

As an open source software work, MiniGUI is mature and successful in technology; the only failure is business: Feynman software has not achieved great development due to MiniGUI. I believe that aside from the business, it is still possible to find a way to keep MiniGUI going.

I have recently concentrated on thinking about how to develop MiniGUI. Later, in the WeChat group of Teacher Song Baohua (a well-known Linux kernel expert), Song teacher asked me for a draft, and I decided the title of this article: "MiniGUI's Nirvana Rebirth Road."

Let's talk about the easiest to talk about: technical aspects

In terms of technology, in the software stack surrounding MiniGUI, we should focus on the following three aspects:

At the bottom, do some work to adapt to the latest hardware and Linux kernel technologies (mainly GPU, OpenGL ES, Open VG, and Direct Rendering Manager). In MiniGUI 3.0, MiniGUI supports virtual buffer rendering in units of processes. By using a specific graphics engine, a single process can be displayed on a separate display surface (Surface). Based on these two points, combined with GPU rendering, you can achieve the UI effect of smartphone operating systems such as Android and iOS.

In order to facilitate the development of MiniGUI applications, it is necessary to develop an integrated development environment, to explore more advanced programming interface packages, and even to use scripting languages ​​to write MiniGUI applications and even general embedded applications. The development of mGNCS and miniStudio during the work of Dong Feifei is a very good attempt and can be used as the basis for continued development. However, due to the development of mobile Internet in recent years, C/C++ developers are becoming more and more scarce. If the new programming framework still uses C/C++ as the programming language, it will have a great impact on the promotion of products or technologies. As a tool, you need to consider lowering the entry barrier for developers. In addition, the new programming interface package should not be limited to GUI programming, but should be integrated with other parts of embedded and IoT device development under a unified application development model. In this regard, the server-side JavaScript framework Node.js developed in recent years is a success story that can be referenced. Of course, the development of embedded or IoT devices has resource constraints. It is difficult to make some trade-offs in terms of development efficiency and program performance. But this may be a place for innovation.

Work with other open source projects with MiniGUI as their native support option, such as WebKit. On the one hand, it can develop key application software on MiniGUI, on the other hand, it can expand the ecosystem of MiniGUI and get more people's attention.

The above-mentioned technical aspects of the unfinished business, as long as there are developers, in time, can be made. But the most troublesome is who will raise these developers, or, where is the money coming from?

If it is a business, what should I do?

After so many years of relying on technology to start a business, I found that it is the most difficult to do business by technology alone. I have to be saddened by any traditional business model, such as opening a restaurant, acting as an agent, and so on. Everyone knows that in the Chinese market, the value of software is often underestimated. The era of making money by selling copies of software is gone forever, so China has never spawned a giant like Microsoft. In such a big environment, Feiman can rely on MiniGUI for a single business for more than ten years. After MiniGUI stops updating, there is still some income, which can be said entirely because of luck. But everything has a double-sidedness. The exploration of Chinese companies in the Internet business model has promoted the rapid development of enterprises such as Tencent and Ali. So, here I can explore a hypothesis: If I can come back, what kind of model would I choose to do the MiniGUI business?

First of all, I will not naively believe that the MiniGUI business can achieve hundreds of millions of annual revenues. I think the annual revenue of 10 million yuan is good for such a software.

Secondly, I believe that the dual-licensing business model adopted by Feynman Software before 2010 was correct. It was the correct choice under the conditions at that time and accumulated funds for the further development of Feynman Software. But in 2010 or earlier, business model adjustments should be made (of course, if I don't think that the life cycle of MiniGUI is over, this is a limitation of the eye).

Finally, around the business model of MinIGUI, the correct approach should be: use the advantages of open source software, continue to expand the user (developer) group, maintain a small-scale (10 to 15 people) software company, to ensure that the company can survive for a long time. In this case, wait for the market to change, look for opportunities to break out or be acquired.

As far as my current situation is concerned, my energy is not allowed to form a complete team to operate MiniGUI and do it as a business. But here, I can give a business model for reference. Of course, this business model is also available for reference to basic software startup teams such as RT-Thread and SylixOS:

Use Apache licenses to release basic software (such as operating systems, libraries, etc.) and do not use GPL/LGPL and other terms. This will allow you to quickly get the most out of your business applications and increase your market share. Be aware that if Android is released under the GPL license, it is absolutely impossible to achieve such great success.

Provides development tools for a fee, such as MiniStudio's integrated development environment, miniStudio, and provides licenses based on the number of developers and years of use. Open source tools are not very meaningful, but such tools run on desktop computers, so you can control their licenses through the cloud, and you can distribute some paid modules or provide technical support services through the cloud. In this mode, the cost of selling the tool software becomes very low, and the user can even place an online order on the webpage and obtain a license by mail.

If we adopt the above business model and develop and maintain 500 paid developers globally through appropriate promotion, we can basically support a small software company of 15 people. If the promotion is effective, it is possible to reach an annual sales scale of 10 million yuan.