Compiling Pattern Matching to In-place Transformation
In this talk, I present a joint work with Laure Gonnord and Gabriel Radanne about compiling pattern matching to in-place transformation. We define a bare-bone high level domain specific language to express in-place modifications of data-structures. This language then serves as our basis to analyze how the in-place modifications can be made by computing conflicting memory regions. We can then compute the order in which the memory movements have to be done. Once this order is known we can then proceed to generate code that will perform the transformation. The main goal of this work is to keep the high-level view provided by pattern matching when dealing with data-structure modifications without the need to create a new instance of the data-structure.
Speaker: Paul Iannetta
Google Meet: https://meet.google.com/myu-dhmz-gvu