Relates to: #12
The JSON generator uses a manual recursion (with the help of the custom type Stack).
@da2ce7 made a proposal using classical recursion here.
We could try to refactor the generator and compare both implementations.
I have the feeling that classical recursion would increase readability, and I don't think that would have a big impact in performace. I used the Stack becuase:
- I decided to port the C implementaion first without making changes and refactor later after adding tests.
- My first attempt was not succesfull. In fact, I started from scratch with classical recursion but without separating tokenizer from generator (parser). It was a mess and I decided to go with the port from C and refactor later.
If we implement this, I think we would have a 100% classical approach: tokenizer + parser with recursion. I think that's good because I think it's a common pattern. On the other hand, I think it would be even easier to add generators for other formats.
We can also continue with the current version until we add other formats. I don't think the Stack version is very hard to follow.
cc @da2ce7 @magecnion
Relates to: #12
The JSON generator uses a manual recursion (with the help of the custom type
Stack).@da2ce7 made a proposal using classical recursion here.
We could try to refactor the generator and compare both implementations.
I have the feeling that classical recursion would increase readability, and I don't think that would have a big impact in performace. I used the
Stackbecuase:If we implement this, I think we would have a 100% classical approach: tokenizer + parser with recursion. I think that's good because I think it's a common pattern. On the other hand, I think it would be even easier to add generators for other formats.
We can also continue with the current version until we add other formats. I don't think the
Stackversion is very hard to follow.cc @da2ce7 @magecnion