Having tries to target two of these environments at the same time I can safely say the if you have to use a database etc. you end up having to write unique code for that environment. Have you got a great way to handle this situation?
Add a comment
|
6 Answers
If you do find you need to write unique code for an environment, use pythons
import mymodule_jython as mymodule
import mymodule_cpython as mymodule
have this stuff in a simple module (''module_importer''?) and write your code like this:
from module_importer import mymodule
This way, all you need to do is alter module_importer.py per platform.
3 Comments
minty
So Daren, is there a way to make module_importer import the right variable dynamically based on some environment variable etc.
Daren Thomas
Sorry, I am drawing a blank here. This would be an excelent SO question though: "How to dynamically find out if we are running Jython, CPython or IronPython?". Then, you could just do it in ifs...
jfs
Python's module
os can be used as an example. It provides OS routines e.g., os.unlink, os.rename, etc depending on what system you're on.@Daren Thomas: I agree, but you should use the platform module to determine which interpreter you're running.
1 Comment
Cyberdrow
As far as I know 'platform' is not available in IronPython (2.0.1). 'os' isn't either.
I write code for CPython and IronPython but tip should work for Jython as well.
Basically, I write all the platform specific code in separate modules/packages and then import the appropriate one based on platform I'm running on. (see cdleary's comment above)
This is especially important when it comes to the differences between the SQLite implementations and if you are implementing any GUI code.
Comments
The #1 thing IMO: Focus on thread safety. CPython's GIL makes writing threadsafe code easy because only one thread can access the interpreter at a time. IronPython and Jython are a little less hand-holding though.
Comments
I'm pretty sure you already know this but unfortunately Jython can't load c extension modules.
Comments
There are two major issues at play here...
Firstly, to my knowledge, only CPython has RAII - you have to close your own resources in Jython, Ironpython, etc.
And Secondly, as has been mentioned, is thread safety.
