#DATA STRUCTURES IN COMPILER DESIGN HOW TO#
#DATA STRUCTURES IN COMPILER DESIGN CODE#

#DATA STRUCTURES IN COMPILER DESIGN HOW TO#

I want to show you how to effectively implement this pre-processing phase. I want you today to try and understand how to recognize this pre-processing stage inside your programs. Then, there's the actual processing phase, which is the execution phase.

The goals for today - our programs, even if we sometimes do not really realize, have often a pre-processing phase when you prepare them for execution. You get other benefits as well for tooling that you might be using, and you will probably might want to use soon. Moreover, if you get a better understanding of how compiler works, you get a better understanding of how tools like GraalVM and Quarkus work inside. Plus, learning about language implementation can give you a very different angle to look at your problems, and so it opens to you different solution spaces. Actually, it's something that all of you can understand here, and probably you might want to use it, or you might be able to use it in your own programs. In fact, the design patterns that we use to design and implement programming languages are pretty simple.

Language implementation is often seen as some dark art, a dark magic, something that only very skilled people can manage to understand. I started to think, 'What if I could solve this problem in a way that it's similar to how programming languages are implemented?' I started to realize that maybe this kind of approach could have been useful for other people too. Even though the two data models are basically very similar, we can not just do trivial mapping. When you read an XML file, what you get it's a tree, while the thing at the top, it's a diagram, so it's not the exact same representation. Basically, it should have been trivial, but in fact, it was much more challenging that you could have imagined. It was presented to me as a data model mapping problem.

#DATA STRUCTURES IN COMPILER DESIGN CODE#

It was one of those tasks that nobody wants to do, because it was a legacy piece of code that reads basically an XML file, a BPMN diagram, which is the thing at the top, for our diagramming tool. My first task at Red Hat was boring at the beginning. That's why I was told so many times, 'Your talks, your papers have to start with proper motivation.' So, hereвЂ™s motivation. Because I did my PhD at University of Milan, even though I left academia, it kind of stuck with me. Right now, I'm at Red Hat, where I work mainly on Drools and jBPM. My name is Edoardo Vacchi A couple of words about me - that's my handle, you can find me on Twitter and GitHub with that handle.ДЄfter doing research at University of Milan where I did mainly programming languages, research, framework stuff, or implementing compilers, interpreters, then I went to UniCredit - which is a large Italian bank, it's a famous bank in Europe as well - in the research and development department, when I did a little bit of research for languages for distributed computation, but mainly I worked on a streaming engine. Vacchi: My presentation is 'Your Program as a Transpiler.' First of all, a couple of words about who I am.

Data structures in compiler design