Learning more about the region-based memory management

[bentxt]

Hi,

I try to learn more about the region based memory management functionality of mlkit. How and when it can be used. So while looking for hints in this codebase I stumpled upon this README in the 'test_dev' directory.
I wonder if this could give some hints on how it is used and where its limitations are.

test_dev/README, there is this note:

Programs that work with the KAM backend:

 a.pm                                 ok
 b.pm                                 ok
...

I read this term KAM (Kit Abstract Machine) somewhere in an older paper. Has this still some meaning? What does those example show excactly?

Ben

[bentxt]

small note: I think this is also interesting #34

[melsman]

To get started with understanding the region based memory management scheme used in the MLKit, please consult the mlkit manual. The test_dev folder is only relevant for compiler development and the pm-files are renamed into mlb-files. Also notice that the KAM backend (i.e., the bytecode backend) is no longer supported, which means that the KAM (Kit Abstract Machine) concept is purely conceptual.