There was a need for controlling behaviour in a Java library project that fell between a JSON configuration file and full blown code.
While Jython, Lua or some similar library could be used it would require version coordination if the app already used those libraries. Additionally those libraries are orders of magnitude larger than the original project which is very undesirable.
Jia Mini Python is just over 800 lines of code (Java)/1,000 (Objective C), and a single source file which you are encouraged to put into your own packages thereby avoiding having to coordinate versions.
Yes. It is used to implement the Appington SDK. The same Python code is used to control audio downloads and playback. For example when playing an audio prompt the locale, time since last audio, caps, and other criteria need to be taken into account.
The SDK is used in several different Android and iOS applications, covering millions of users all over the world.
Code run by Jia Mini Python is restricted in what it can do. For example it can only access modules you explicitly make available, and then only public methods implemented directly in the module.
You can return and accept any objects, but the code cannot access them other than passing them around. This means you could expose instances but do not have to worry about methods being called on the instances that would put them into undesirable states.
Code can perform denial of service attacks:
# infinite loop
while True:
pass
# huge list
l=[]
while True:
l.append(l)
# huge string
s="a"
while True:
s=s+s
Consequently you should not use “untrusted” code.
There is a comprehensive test suite. Since Python already exists it provides something to verify answers against. The test harness is written in Python.
There is over 99% line coverage and similar branch coverage in the code. While this doesn’t guarantee the absence of bugs, it at least assures that the vast majority of code is tested. (The remaining lines/branches are generally defensive coding.)
The scope being limited (note “Mini” in the name!) there is also less functionality and less to go wrong.
Please use the discussion group https://groups.google.com/forum/#!forum/jia-mini-python and bug tracker at https://github.com/rogerbinns/jia-mini-python/issues if you encounter any issues.
There is no debugger provided or any similar notion of interactivity. It is easiest/quickest to develop your code using print statements. On a host computer you can use normal Python mocking the modules.
When there are errors you get a line number that can be looked up, as well as a program counter which can be used with an annotated listing to find the exact issue.
Use asserts to verify operation of your code. By default they are ignored but you can turn them on passing --asserts to jmp-compile. A large portion of the test suite is implemented using asserts.