2016 year-end summary

Sort things out this year:

Job:

1. Im communication Platform
Completely redesigned from Erlang and written from scratch. Main functions, multi-terminal message, read unread, History full synchronization, Rich text, friend relationship, group, discussion group, two people, multi-person audio and video. Enable flexible automated deployment, log collection analysis, monitoring alarms, and full-use case coverage of automated test scripts.
Invested 120% of the energy. The SDK section benefits from the LUA implementation and code generation scripts, with relatively low development effort, and even with the Android/ios/windows/nodejs (WINDOWS&MAC)/WEBJS 5 platform versions, only 1 platform scripting costs are required.
The first 5 years have been engaged in the development of IM backstage, but more only in the careful maintenance of a lump of "old guy", heavy in seeking stability. Finally have the opportunity to start from the beginning, the upfront spend a lot of energy from the analysis of Research, Telegram, WhatsApp and other IM protocol design, to find technology sharing, analysis conjecture server design. Lead the redesign of the Protocol, platform architecture, find ways to make things the simplest is the way elegant and beautiful implementation. Not good, also dare to push to redo, group protocol because there are always some awkward, platform and test scripts are almost three times again. A good design must be aesthetically pleasing, and nothing can be done.
Codereview use of ph review after the submission method, of course, the initial may not be very happy, but adhere to a period of time after the effect is really good. I can't imagine how bad the code would be without review. Start to see review may take up 30%+ time, a lot of comments repeatedly modified, read the code can know a person's development level, character, advantages and disadvantages, there is no pursuit, of course, as far as possible to remind improvement, want to improve only a lot of time just wishful thinking, finally can arrange suitable for the right things.

2. Audio and video engine
Android/ios/windows three platforms, Android down very familiar, IOS and Windows completely novice. It's a new field for yourself, c++/oc/ndk/opengl/mfc/d3d .... Take up less than 30% energy, live carrying almost one months to WEBRTC overall process and each platform probably whole understand. The main development work is feature, such as the implementation of CVO (coordination of video Orientation), adjust the quality of videos, iOS Videotoolbox hardware encoding support, adjust D3D image or draw, screenshots and so on. But for the real difficulty, the quality of audio calls, after still need to spend a lot of energy to chew down.

3. Other

-Blockchain analysis of bitcoin, Ethereum, IPFs, Lighting network technology implementation

Limited to purely distributed networks in the WAN can only be implemented 10~20 tx/s is not practical, most of the alliance chain application of large institutions between the transaction, such as banks.

-IoT development is simple, chip-fit trouble

- Artificial intelligence is not known for its mature framework, but it's already used

On the whole:

This year from the original pure server, to the end of each platform, including hardware chips are all over, plus audio and video, GPU drawing, deep learning, but also very smooth, technically also can be very good hold, basically no problem left.

The past is a direction of depth, now is the expansion of the breadth of the other side of the world, even in the return to the platform development design, can be more comprehensive assessment of the details, to make a more appropriate plan.

In addition is encountered do not know the technology, not guilty, most of the same, spend some energy can be done, the big deal dead knock, nothing to do.

Learn:

In the first half of the work, from the attention of the Li Yu to the "Road to Wealth Freedom", began to realize that this situation is not, need to set aside some attention.

1. Get: 1. The path to freedom of wealth must be seen, 2. Wan Weigang Elite Day Lesson Basic all listen to 3. The Wang Yuquan outpost is too much to hear.

2. Panda Academy: Subscribe, 1 books per week, start to see the weekend exams every day, now only read the books you like

3. English: First by the word, want to improve English, by some things delayed, next week to start to continue

2017 Highlights:

1. English vocabulary

2. Artificial intelligence, first listen to the course, review mathematics, algorithm

3. The path to freedom of wealth

4. Read more books, all kinds of book

2016 year-end summary