Google launched Python Performance Acceleration Program
 

Yesterday, Google released a Python performance acceleration program - Grumpy, Python code can be translated (transcompile) Go to the source code and then compiled to Native Code. Currently on Github we have received 3000+ star.

Project Address: https://github.com/google/grumpy

Google's move is a Python community has been added to speed up the performance scheme. In many programming languages, Python has been criticized performance is not high, an important reason behind it is its standard implementation of CPython Global Interpreter Lock (GIL) exist, limiting concurrency and parallelism. All along, the industry and the community are looking for a better solution, currently well-known are PyPy, Microsoft Pyjion, and so on Dropbox's Pyston JIT implementation.

Python in Google's widely used, YouTube and the YouTube API front-end server is to use the Python language. Python's father had also Google office.

According to the original intention of the relevant Bowen, this project is to improve the performance of Python code. Grumpy is both a source code translator, but also provides a run-time. Currently, the project is still in the experimental stage, and does not support Python 3.

Responsible for the development of the project engineer to answer netizens questioned why only supported on Hacker News 2:00 Python, he pointed out that "Google currently has a lot of Python 2.7 code base, so is our focus." There is also speculation that Google might move is in response to termination of maintenance related to the time Python 2.7 in 2020. Rather than upgrade code to Python 3, Google will be part of the future are more willing to be translated into Python code Go code.

However, this hypothesis does not seem too much based on. Google engineers made the following response, "the purpose (of the project) is to continue to use the Python code, the code can not be directly translated after secondary treatment. Of course, it may be rewritten for the critical part of the performance with Go. Future actions is a mixed mode. "

Community as well as through the introduction of Rust, Lua, Julia languages ​​such as Python accelerated approach.

Since Python itself has failed to properly address the performance limitations caused by GIL, future community should continue to have other solutions. At present there should be immediate JIT compile the code and mixing in two forms, but also get a lot of big companies.

The trend for future learning Python is also worth noting: the future is certainly not as long as Python will be able to meet the work requirements, you may also need to learn other languages ​​and to study how to Go in combination.