Refactor extension classes

[sgillies]

After #275 is finished, I'd like to try refactoring to eliminate redundant code. Ideally, ogrext1.pyx would have GDAL 1.x compatible Session and Iterator classes that derive from a base classes and ogrext2.pyx would have GDAL 2.x compatible classes that derive from the same base classes.

[snorfalorpagus]

What's the best way to do this? Concatenate an ogrext_base.pyx and the relevant ogrext1.pyx or ogrext2.pyx in setup.py?

[snorfalorpagus]

There is only 1 line different in the writerec method between the two versions. Does this mean the entire function needs to be duplicated? In C you would use an #ifdef but there isn't an equivalent in Cython. Do we just have to put up with this? At the same time I don't want to go down some crazy nonstandard metaprogramming route...

https://github.com/Toblerity/Fiona/blob/master/fiona/ogrext1.pyx#L936
https://github.com/Toblerity/Fiona/blob/master/fiona/ogrext2.pyx#L1000

[micahcochran]

There are two approaches that I can think of. Write a common base class with all the functions that will compile in both versions of GDAL. Then have your GDAL version specific classes derive from a those common base classes. (EDIT: This is what Sean suggested above.)

The other is to write an emulation layer. So, the Fiona code may target GDAL 2.0, but have wrapper functions for GDAL 1.x in the version 2 API's names. This could present a problem in the future if there are functions that Fiona wants to use in GDAL 2.x that aren't easily supported by GDAL 1.x API.

There are a few other ways to do this that add more layers of abstraction. There are probably a few other ways to go. Both of the approaches should reduce the lines of code. Which approach is less confusing?

[snorfalorpagus]

It looks like Cython does support some preprocessor-like functionality. I wasn't aware of this.

http://docs.cython.org/src/reference/language_basics.html#compile-time-definitions

[sgillies]

I'm going to push this one off to a future milestone unless there are objections. It's time to get 1.7 out there!

[snorfalorpagus]

I agree that this shouldn't hold back the release of 1.7 as there is a real need to have GDAL 2.x support now.

[snorfalorpagus]

Closed by #461.